IJlSni/lEItSWnif C)F

GRADUATE COLLEGE

BIOSYSTEMATICS OF NORTH AMERICAN

SPECIES OF (LAMIALES)

A Dissertation

SUBMITTED TO THE GRADUATE FACULTY

in partial fulfillment of the requirements for the

degree of

DOCTOR OF PHILOSOPHY

By

PHILLIP TERRY CRAWFORD Norman, Oklahoma 2003 UMI Number: 3109064

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

UMI

UMI Microform 3109064 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Copyright by PHILLIP TERRY CRAWFORD 2003 All Rights Reserved. BIOSYSTEMATICS OF NORTH AMERICAN SPECIES OF (LAMIALES)

A DISSERTATION APPROVED FOR THE DEPARTMENT OF BOTANY AND MICROBIOLOGY

BY

r. jÆ lisens, CommitteCommittee chair

Ml. Dr. Ola M. Fincke c Slo/arla Dr. Gordon E. Uno

Dr. Linda L. Wallace ACKNOWLEDGEMENTS

This research project was funded through grants provided to the primary author by the University of Oklahoma Graduate Student Senate. Additional research funds were provided through a National Science Foundation grant BSR-8708369 to Dr. Wayne J.

Elisens.

I would like to recognize the members of my graduate committee for their critical reviews of this project. In addition, Dr. Scott Russell made available microscopes and imagingequipment in the Samuel Roberts Noble Electron Microscopy Laboratory; Mr.

Bill Chissoe and Mr. Greg Strout provided advice and assistance in the operation of that equipment. Dr. Gordon Uno provided the use of a controlled-environment chamber and support through summer assistantships. Mr. Dan Hough provided invaluable assistance with computers and software.

IV TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS iv

LIST OF TABLES vii

LIST OF FIGURES viii

LIST OF APPENDICES x

ABSTRACT OF DISSERTATION xi

CHAPTER I. Genetic Variation and Reproductive Biology of North

American species of Nuttallanthus (Lamiales) 1

Abstract 2

Introduction 4

Materials and Methods 7

Results 12

Discussion 19

Literature Cited 31

Tables 38

Figures 53

Appendices 59

CHAPTER n. Morphological Variation and Systematics o f North

American species of Nuttallanthus (Lamiales) 81

Abstract 82

Introduction 84

Materials and Methods 87 Page

Results 91

Discussion 96

Literature Cited 100

Tables 104

Figures 117

Appendix 133

VI LIST OF TABLES

Page

CHAPTER I. Genetic Variation and Reproductive Biology of North

American species o îNuttallanthus (Lamiales)

1. Collection data for populations of three

species examined for isozyme variation 38

2. Genetic identity coefficients among populations 43

3. Summary of allozyme variation 44

4. Nei's genetic diversity statistics 49

5. Observed and expected frequencies of heterozygotes 50

6. Wright’s F statistics 51

7. Intra- and interspecific seed production

following controlled pollinations 52

CHAPTER II. Morphological Variation and Systematics of North

American species of Nuttallanthus (Lamiales)

1. Collection data for populations of three Nuttallanthus

species examined for morphological and

chromosome number variation 104

2. Morphological characters examined 109

3. Descriptive statistics for morphological characters 112

4. Discriminant function analysis of

morphological characters 116

vii LIST OF FIGURES

Page

CHAPTER I. Genetic Variation and Reproductive Biology of North

American species of Nuttallanthus (Lamiales)

1. Collection localities for populations of three

Nuttallanthus species examined for

isozyme variation 53

2. UPGMA phenogram of populations derived

from Nei’s (1972) genetic identity 55

3. Three-dimensional plot of MDS analysis

of Nei’s (1972) genetic identities 57

CHAPTER n. Morphological Variation and Systematics of North

American species of Awrtu/ZuMtAwj' (Lamiales)

1. Collection localities for populations of three

Nuttallanthus species examined for

morphological and chromosome

number variation 117

2. SEM micrograph of whole seed of

Nuttallanthus canadensis 119

3. SEM micrograph of longitudinal ridge and

face on seed of 119

4. SEM micrograph of distal end of seed of

AwiZaZZoM^Aws' 119

V lll Page

5. SEM micrograph of unicellular papilla on

seed of Nuttallanthus canadensis 119

6. SEM micrograph of whole seed of

AWfaZ/anfAity 121

7. SEM micrograph of multicellular tubercles

on seed of Nuttallanthus texanus 121

8. SEM micrograph of whole seed of

Nuttallanthus texanus 121

9. SEM micrograph of longitudinal ridge on

seed of TVwMaZZantAfty teuranwg 121

10. SEM micrograph of whole seed of

yZonWaMw,; 123

11. SEM micrograph of distal end of seed of

123

12. Two-dimensional plot of PC A analysis of 35

reproductive and vegetative characters 125

13. Two-dimensional plot of PC A analysis

of 26 reproductive characters 125

14. Two-dimensional plot of PC A analysis

of 9 vegetative characters 127

15. Two-dimensional plot of PCA analysis of

population means 127

IX Page

16. Three-dimensional plot of PC A analysis of

population means of 35 reproductive

and vegetative characters 129

17. UPGMA phenogram of populations derived

from average taxonomic distance 131

LIST OF APPENDICES

CHAPTER I. Genetic Variation and Reproductive Biology of North

American species of Nuttallanthus (Lamiales)

1. Diagrams of composite banding patterns of

alleles of scored loci 59

2. Genotype data for loci from 60 populations of

three species of AwfW/aMfAwj: 61

3. Summary allele frequencies for 14 loci in

three species of Vw/ta/ZantAwi' 75

4. Fixation indices for 14 loci in 42 populations

of three species of 77

CHAPTER n. Morphological Variation and Systematics of North

American species of Nuttallanthus (Lamiales)

1. Eigenvalues and loadings from principal component

analysis of morphological variation 133 ABSTRACT OF DISSERTATION

The native North American toadflaxes comprise a morphologically variable complex of herbaceous plants that have been collectively recognized as a section of

Linaria (Linaria section Leptoplectron) or a distinct genus {Nuttallanthus D. Sutton) and individually assigned to specific or varietal rank. High levels of intraspecific morphological variation and intergradation in the morphological characters used to distinguish among these taxa have led to diverse taxonomic treatments of the members of this group. This study represents the first use of enzyme electrophoresis to examine patterns of genetic variation in this species complex. Despite their wide geographic

distribution and showy, fragrant flowers (seemingly adapted for insect pollination and

outbreeding) the extent and distribution of genetic variation at the populational level was

similar to that of narrowly distributed, autogamous species; a low level of detected heterozygotes suggested a high degree of inbreeding in these taxa. Field and greenhouse

breeding-system studies indicated that individuals of all three species were entirely self­

compatible and that the primary reproductive strategy of these taxa involved cleistogamy

and self-pollination. No hand-performed cross pollinations among species resulted in the

successful development of capsules and seeds, indicating that these species were

reproductively isolated in areas of sympatry. Univariate and multivariate analyses of

morphological characters measured from field-collected samples revealed significant

morphological gaps among taxa and consistently separated samples into three primary

clusters corresponding to the currently circumscribed species; strong and significant

correlations between patterns of morphological and genetic variation were observed

among populations and species. The observed morphological discontinuities separating

XI groups in this genus and the close correspondence of extensive morphological and genetic gaps among these taxa are concordant with the observed cross-incompatibility among species; these results strongly support the recognition of three species in North

American Due to similar genetic divergence values among species, phylogenetic relationships in Nuttallanthus remain unresolved. Patterns of genetic and morphological differentiation within and among these species suggests a long history of reproductive isolation due to biotic and paleogeographic barriers to gene flow.

xn CHAPTER 1.

GENETIC VARIATION AND REPRODUCTIVE BIOLOGY OF

NORTH AMERICAN SPECIES OF (LAMIALES) ABSTRACT

The native North American toadflaxes comprise a morphologically variable complex of taxa that have been collectively recognized as a section of linann (Zmonn section or a distinct genns D. Sutton) and individually assigned to specific or varietal rank. To assess patterns of genetic variation within and among species, starch gel electrophoresis was employed to examine 15 scorable isozyme loci for 50 populations. Species are widely distributed geographically, occur in structurally-homogeneous populations ranging in size from one to thousands of individuals, and possess showy, fragrant flowers seemingly adapted for insect pollination and outbreeding. Despite these characteristics, the amount and apportionment of genetic variation within and among populations was similar to that of narrowly distributed, autogamous species. A low level of detected heterozygotes suggestedhigh a degree of inbreeding. Genetic identities among species were quite low (ranging &om 0.516 to

0.623) relative to intraspecific identity values (ranging from 0.819 to 0,936); qualitative differences among species were evident at several loci. Field and greenhouse breeding- system studies indicate high levels of self-fertilization; all individuals studied were entirely self-compatible and many chasmogamous flowers selfed prior to anthesis.

Individuals of all three species commonly produced cleistogamous (obligately-selfmg) flowers both early and late in the life cycle. No hand-performed cross pollinations among species resulted in the successful development of capsules and seeds. The low interspecific genetic identity estimates, high proportion of unique alleles (ranging from 4 to 9 per species), and observed cross-incompatibility among taxa support recognition of three North American species in Nuttallanthus. The extent and apportioning of populational genetic variation within these taxa, uniformly low numbers of heterozygotes, and high degree of association between genetic distance among conspecific populations and geographic location are concordant with a mixed mating system characterized by autogamy and facultative xenogamy. INTRODUCTION

The genus D. A. Sutton was established and segregated &om

Zmana Miller in 1988 and currently includes one South American species (A.

(Diels) D. A. Sutton) and three species of native North American annual or biennial herbs: N. canadensis (L.) D. A. Sutton, N. floridanus (Chapman) D. A. Sutton and A. teroMWJ (Scheele) D. A. Sutton (Sutton 1988; USDA/NRCS 2002). These plants possess heteromorphic stems and produce showy personate, bilabiate flowers with anterior nectar-storing spurs or pouches that attract a variety of Lepidopteran and

Hemipteran visitors. Generic segregation as well as specific and varietal delimitation in this group of plants have been made primarily on the basis of variation in corolla morphology, spur length, and seed coat surface ornamentation. While the variation in floral morphology among taxa has been attributed to adaptation to different categories of pollinating insects, with N. canadensis and N. texanus attracting butterflies and N. floridanus being pollinated by flies (Pennell 1935), no systematic study of the mating system or pollination ecology of these species has been reported. These species commonly grow in dry sandy soils of dunes and open coniferous woodlands and as weeds of fields and other heavily disturbed areas. and A. tezuMWj' possess the greatest geographical ranges of any species among New World Antirrhineae

(Elisens 1985). Nuttallanthus canadensis is native throughout much oftemperate North

America and is naturalized in South America and in Europe, where it has been cultivated as an ornamental for its showy, fragrant flowers (Tutin et al. 1972). AwtmZZaMtAws'

is native to the southern United States and Mexico, may be native to temperate

South America, and is naturalized in other temperate regions (Sutton 1988). yZonWa/zwj is more narrowly distributed and occurs in the Atlantic and Gulf coastal plain of a few states in the southeastern U.S.A. Species ranges overlap, and all three taxa occur occasionally in dense, mixed populations.

The type species of the genus (N. canadensis) was described originally by

Linnaeus (1753) as vfrnizfrAmuM L. All native North American members of tribe Antirrhineae with spurred corollas were quickly reassigned to the genus Linaria by

Miller (1754; 1768) andZ. (L.) Dumort. was recognized in 1802 (Dumont de

Courset 1802). Lmuna ietuMu Scheele (1848) andZ.yZontfuMa Chapman (1860) were originally described as distinct species, but Pennell (1920, 1922) recognized only L.

and L. /ZonWaMU in his North American Lmunu section Zeptpp/ecP-oM.

Linaria canadensis was composed of two varieties: L. canadensis and£. canadensis var. texana (Scheele) Pennell that exhibited a high degree of intergradation in flower size.

Munz (1926) concurred with Pennell's (1920) delimitation o ff. and suggested that the intergradation in floral features between the two taxa was “quite complete” and that they could be distinguished reliably only on the basis of seed coat morphology. In 1935, Pennell reconsidered his earlier opinion and recognized L. canadensis and L. texana as distinct species, stating that the reported intergradation in flower size and seed structure could be attributed to ecological factors and hybridization among species. Several subsequent treatments relegated L. t&ru/za to varietal rank as L.

var. texuma (Scheele) Pennell (Rothmaler 1954; Cronquist et al. 1984,

Hitchcock et al. 1998). In addition, plants possessing cleistogamous flowers and white corollas have been accorded infraspecific rank as 1. forma cZeij'togumu

Fernald and 1. canadensis forma albina Femald, respectively (Femald 1936; Femald 1943). In a recent treatment of tribe Antirrhineae, Sutton (1988) recognized three native

North American species in AwtmZ/uMtAw.; distinct 6om the Eurasian species of Zmonu.

Specific delimitations were made primarily on the basis of differences in seed coat

structure, pedicel length, and corolla size. The extent of morphological variation within

species and potential interspecific hybridization among species were not addressed

(Sutton 1988).

Although species boundaries and evolutionary relationships are unresolved among

New World toadflaxes, previous morphological studies supported hypotheses that the

complex forms a monophyletic group distinct from the 150 Eurasian species o f Linaria

and from other New World members of tribe Antirrhineae. The size, surface

morphology, and testai anatomy of the seeds of Nutm/ZuntAug species are unique among

New World Antirrhineae (Elisens et al. 1983; Elisens 1985), as is the floral structure

(Rothmaler 1943), chromosome base number of x = 6, and pollen morphotype (Elisens

1986). In addition, N. canadensis possesses tubular nuclear inclusions in leaf mesophyll

cells that are otherwise unique toLinaria among genera in Tribe Antirrhineae (Bigazzi

1989; Bigazzi 1993). Although pollen size and exomorphology of Nuttallanthus is

similar to that of examined species of Imun'u (Elisens 1986), the floral structure and seed

coat morphology and anatomy of AwttuZZuMtAMj' suggested that the New World species are

taxonomically and phylogenetically distinct 6om Old World Zmunu (Elisens et al. 1983;

Elisens 1985; Sutton 1988).

This study employed starch gel electrophoresis of soluble enzymes and hand-

pollination experiments to document the genetic variation and reproductive biology of the

North American species of The principal goals of this investigation were to determine the amount of genetic differentiation within and among species in the genus,

to test hypotheses of species boundaries and relationships, and to document the types of

mating systems and the degree of reproductive and genetic isolation exhibited by these

species.

MATERIALS AND METHODS

awafygw. A total of 649 individuals &om 50 populations representing 3

species of Nuttallanthus were examined for electrophoretic variation; 325 individuals

from 22 populations of N. canadensis, 110 individuals from 8 populations of A. floridanus, and 214 individuals from 20 populations of N. texanus (Table 1; Figure 1).

Voucher specimens of the populations sampled were deposited at the Bebb Herbarium of

the University of Oklahoma (OKL). Individual plants were collected by walking a

transect along the greatest dimension of a population and collecting entire plants at

intervals selected to attain a sample size appropriate for the size of that population; a

maximum of 20 percent of any given population was sampled, although this limited the

sample size obtainable from small populations. Individuals of populations 42 and 43

consisted of plants grown from field-collected seed. Plants were bagged individually and

placed in plastic containers on ice for transport to the lab, where they were refrigerated at

4°C prior to protein extraction. Tissues hom young, actively-growing leaves served as

the enzyme source.

Fresh leaf material was ground in an extracting buffer consisting of 0.1 M tris-

HCL pH 7.5, 1 mM EDTA (tetrasodium salt), 10 mM MgCLz, 10 mM KCl, 14 mM beta-

mercaptoethanol and 20 mg/ml solid polyvinylpyrrolidone (following Gottlieb 1981b). Leaf extracts were centrifuged and the supernatant was absorbed onto wicks of Whatman

17 MM chromatography paper, which were then stored in 1.5 mL microcentrifuge tubes at -70°C for approximately 1 hour. Samples were electrophoresed on 11% starch gels using two buffer systems (Soltis et al. 1983) to resolve 15 loci for 10 enzyme systems: aspartate aminotransferase (AAT), alcohol dehydrogenase (ADH), CT-glycerophosphate dehydrogenase (GPD), isocitrate dehydrogenase (IDH), malate dehydrogenase (MDH),

6-phosphogluconate dehydrogenase (6PGD), phosphoglucoisomerase (PGI), phosphoglucomutase (PGM), superoxide dismutase (SOD) and triosephosphate isomerase (TPI). System I consisted of an electrode buffer of 0.18 M Tris and 0.004 M

EDTA titrated to pH 8.6 with boric acid and a gel buffer prepared from a 1:3 aqueous dilution of the electrode buffer; System I was used to resolve AAT, ADH, GPD, PGI,

PGM, SOD and TPI. The electrode buffer of System II was prepared from 0.065 M L- histidine free base titrated to pH 6.5 with citric acid monohydrate and a gel buffer was obtained from a 1:3 aqueous dilution of the electrode buffer; System II was used to resolve IDH, MDH and 6PGD. Agarose-overlay and staining procedures used to detect enzyme activity followed the protocols of Soltis et al. (1983). The genetic bases of the enzyme banding patterns observed were inferred from their concordance with published data regarding the basic number of loci expected in the absence of gene duplication and patterns of enzyme expression (Crawford 1990; Gottlieb 1982). The documented number of independent banding regions and the patterns of banding within those regions were consistent with the known substructure and compartmentalization of the resolved enzymes. Where more than one locus and allele were observed, loci were numbered and alleles were lettered beginning with the most anodal (fastest-migrating) form. Allele and genotype frequencies were determined for each population and species.

The percentage of polymorphic loci, mean number of alleles per locus, mean number of alleles per polymorphic locus, mean observed heterozygosity, Nei's (1987) gene diversity, and the effective number of alleles per locus were calculated manually for each population and were averaged across all populations of each species. In addition, species-level statistics were computed by treating all sampled individuals within each species as members of a single population (following Hamrick et al. 1990); thus, the average genetic diversity of populations could be compared with the degree of genetic diversity found within each species. Spearman’s rank-order correlation analysis was performed with SPSS for Windows (ver. 11.5.0; SPSS, Inc. 2002) to examine the association of population size with the genetic diversity of populations.

The total genetic diversity (Hr) within each species was partitioned into within- population (Hg) and among-population (Dgr) components (Nei 1973); the proportion of genetic diversity among populations (GgT) and Wright’s (1951) estimate of gene flow

(Nm) were calculated for each species. To further assess the level ofgenetic differentiation among populations of each species of Wright’s F statistics

(Fis, Frr and Fgi) were computed for each polymorphic locus and were averaged across loci within each species. Wright’s fixation indices (F) were calculated for each locus and population of the three species of the statistical significance of observed deviations of heterozygote proportions from Hardy-Weinberg expectations was assessed using a Chi square analysis (Wright 1965,1978; Nei 1977). N ei’s (1972) genetic distance and genetic identity values were computed for all pair-wise combinations of populations and species. Roger’s distance as modified by Wright (1978) and Cavalli- Sforza and Edwards (1967) chord distance coefficients were calculated and were used to produce dendrograms using the Distance Wagner procedure (Farris 1972); cophenetic correlations were calculated for each of the dendrograms. BIOSYS-1 (Swofford and

Selander 1989) and GENESTAT2 (Whitkus 1988) provided computational software.

NTSYSpc (ver. 2.11c; Rohlf 2002) was used to calculate Nei's (1972) genetic identity and distance, Roger’s distance as modified by Wright (1978) and Cavalli-Sforza and

Edwards (1967) chord distance for all pairwise comparisons among populations and species. Dendrograms were generated using these coefficients with the unweighted pair- group method using arithmetical averages (UPGMA) as discussed in Sneath et al. 1973; cophenetic correlations were calculated for each of the dendrograms. NTSYSpc was also used to conduct nonmetric-multidimensional scaling (MDS - Kruskal 1964) of Nei’s

(1972) genetic identities among populations. Species-level genetic diversity statistics and genetic distances and identities were calculated using all loci; only polymorphic loci were included in calculations of genetic diversity within and among populations, because statistics excluding monomorphic loci better represent the partitioning of genetic variation within and among populations (Berg and Hamrick 1997).

To assess the degree of association of the genetic distance between populations within each species and the geographic location of those populations, Spearman's rank- order correlation analysis was perfbmed with SPSS for Windows on matrices of Nei's

(1972) genetic distances and of linear distances (in kilometers) between populations.

Crossing studies. A minimum of five individuals was propagated from seed collected from each of 25 populations representing three species of AwtmZZaMtAw.;. Due to the erratic germination of untreated seeds, seeds were soaked in a solution of gibberellic

10 acid (O.lg / L) for 10 minutes, washed with distilled water and germinated at room temperature on water-saturated filter paper in parafilm-sealed Petri dishes. Seedlings were transplanted to a well-drained soil-based potting soil mix and cultivated under pollinator-&ee conditions in a Conviron CMP2023 growth chamber (13.3 hrs of light at

28°C; nights at 16°C). The crossing program involved tests for autogamy (observation of untreated cleistogamous and chasmogamous flowers), apomixis (emasculation of flowers followed by no hand pollination), self-compatibility (emasculation followed by geitonogamous hand pollination) and cross-compatibility (emasculation followed by intra- and interspecific hand pollination). Buds of manipulated flowers were emasculated with needle-point forceps prior to anthesis, because the anthers of cleistogamous and chasmogamous flowers commonly dehisced prior to the opening of the perianth. Hand pollinations were accomplished one day following emasculation; forceps sterilized in

95% ethanol were used to transfer recently dehisced anther sacs to the receptive stigmas of emasculated flowers. Flowers used in reproductive experiments were not bagged. Due to the small size of the pre-anthesis floral structures, all floral manipulations were performed utilizing an Olympus dissecting microscope at 12.5X.

Two hundred forty hand pollinations representing all possible directional interspecific crosses were made; a minimum of 30 and a maximum of 50 hand pollinations were made for each directional interspecific combination. In addition, 75 or

80 intraspecific hand pollinations were performed among populations of each species;

100 hand pollinations using self pollen were performed for each species.

Seed production and germination rate were recorded for 100 capsules produced by both cleistogamous and chasmogamousflowers of each species and for all of the

11 manipulated flowers that produced mature capsules and seeds. Seeds were counted under a Wild M5 stereo microscope at 12X. 25 seeds from each capsule were soaked in a solution of gibberellic acid (O.lg / L) for 10 minutes, washed with distilled water and germinated at room temperature on water-saturated filter paper in parafilm-sealed Petri dishes.

Univariate statistical analyses were performed with SPSS for Windows (ver.

11.5.0; SPSS, Inc. 2002). Descriptive statistics of seed production and germination rate were calculated for each species; normality tests were performed for each treatment category. To compare treatment categories among populations and species, one-way analysis of variance was employed; post-hoc testing was performed using Fisher’s least significant difference test.

RESULTS

Isozyme analysis. Fifteen loci coding for 10 enzymes were scored from populations of three species of Nuttallanthus: three for MDH, two for AAT, PGI and TPI, and one for ADH, GPD, IDH, 6PGD, PGM and SOD (Appendix 1); MDH-1 and MDH-2 overlapped so extensively that the loci and alleles could not be reliably distinguished; three banding patterns were observed, so MDH-1 and MDH-2 were treated collectively as a single polymorphic locus with patterns A, B and C corresponding to alleles

(following Small et al. 1999). Additional isozymes were detected for AAT, ADH, IDH,

6PGD and PGM but were not scored due to low or inconsistent levels of activity or poor resolution. Additional enzyme systems detected included ME (using buffer system I);

ALD, GA3PD, ME, MNR, and SDH were detected on buffer system II; these enzymes

12 were not scored due to low or inconsistent levels of activity or poor resolution. The number of isozymes detected was typical for diploid plant species, with the exception of

MDH, which generally has three loci, and PGM, which normally has two (Gottlieb 1982;

Weeden and Wendel 1989); no differences in isozyme number among species of

was observed. The number of allelomorphs observed at polymorphic loci ranged &om two (AAT-1, MDH-4, SOD) to five (AAT-2). One isozyme (MDH-3) was

invariant in all examined individuals. Most of the populations of Nuttallanthus texanus

examined in this study were composed of individuals possessing one of two principal

allozyme profiles; individuals possessing different profiles were also distinguishable on

the basis of seed coat morphology (Crawford 2003). One population in central Oklahoma

(population 58) was composed of individuals belonging to both of these major groups of

N. texanus-, that population was divided into two sub-populations (58a and 58b) for

analysis. Genotype data for populations are presented in Appendix 2; summary allele

hequencies for three species of Awtta/ZuMfAwj' at 14 polymorphic loci are presented in

Appendix 3.

Mean values for Nei’s (1972) genetic identity coefficients (I) for pairwise

comparisons of 50 populations within and among AwtraZ/aMt/zw.; species are presented in

Table 2. Average I values within species varied from 0.819 (A( /e%anw.y) to 0.936 (#.

cuMudeMj'i.y). Mean genetic identity values were considerably lower between species than

within species, with interspecific I values ranging from 0.516 (A. cawadeMJM x A. floridanus) to 0.623 {N. canadensis x N. texanus). Comparable patterns of genetic

similarity were observed regardless of the coefficient employed (including Roger’s

distance as modified by Wright (1978) and Cavalli-Sfbrza and Edwards (1967) chord

13 distance). The average of all pairwise comparisons among Nuttallanthus species (I =

0.580) is considerably lower than the average identity value (I = 0.670) for plant congeners reported by Gottlieb (1981a).

The observed pattern of genetic similarities among species of Nuttallanthus results 6om both qualitative and quantitative allelic differences between those species.

Many populations were fixed for a single allele at a majority of loci. The three species

shared the same highest-frequency allele at 6 of 14 polymorphic loci. Of the 46 alleles

observed among 14 polymorphic loci, 19 were unique to one of the three species; 3

“marker” alleles were either fixed or were present in high frequencies in all populations

of a single species (6PGD-2c and TPI-ld in M and 6PGD-2b in M yZondaMW.s). and #. fexanwj' had the greatest number of alleles in

common (20 of the 31 alleles detected in N. texanus were shared with N. floridanus)-, N.

texanus shared an equal number of fixed or high-frequency alleles (9 of 24) with N.

and #. yZonzfa/nty, whereas TV. and jV. shared 6 fixed or

high-frequency alleles.

Population- and species-level genetic diversity estimates based on allele

frequencies are provided in Table 3. Within populations, the mean number of alleles per

locus (A) ranged from 1.00 to 1.64; the mean number of alleles per polymorphic locus

(ApJ varied from 2.0 to 3.0; and the percentage of polymorphic loci (F) ranged from 0.0

to 57.1. The average proportion of heterozygous loci perindividual (Ho) ranged from 0.0

to 0.054; the average proportion of heterozygous loci per individual expected for

populations in Hardy-Weinburg equilibrium(Hg) ranged from 0.0 to 0.283. Populations

of N. ca/zuzfgMj'zj' exhibited the highest mean values of ^ (1.25), f (23.7) and Tfg (0.076),

14 whereas #. populations showed the lowest levels o f genetic variation among the three species. Spearman’s rank-order correlation analyses indicated positive and statistically-significant associations between population (sample) size in Nuttallanthus andv4 (r^ = 0.50, p < 0.001), f (r^ = 0.52, p < 0.001),(r, = 0.33, p = 0.019), andf^fg (r^

- 0.49, p< 0.001).

Estimates of genetic variation within populations of Nuttallanthus species were relatively low in comparison to values reported for other species of Scrophulariaceae

(Elisens et al. 1988) and were similar to those reported for populations of selling species and endemic species with narrow geographic ranges (Hamrick 1989; Hamrick et al.

1990). The effective mean number of alleles per locus (Ag) within populations varied from 1.00 to 1.39. Ag is equal to the actual number of alleles only when all alleles exist in equal frequencyand provides a measure of allelic evenness; this value was lower than the mean number of alleles per locus (A) observed in all polymorphic populations, suggesting that a portion of the allelic diversity within populations was present in the form of low-

&equency alleles (Nei 1987).

Genetic diversity within the three species of Nuttallanthus was considerably higher than that observed within populations of those species, with Æ rexonw.; showing values o f (2.21), f (71.4) and (0.234) equal to or higher than those of M caMacfema'w;

#. exhibited the lowest levels of genetic variation among the three species

(Table 3). The total genetic diversity within each species (HT), mean genetic diversity within populations (H§) and among populations (Dsi), the proportion of genetic diversity among populations (GsT) and Wright’s (1951) estimate of gene flow (Nm) are presented in Table 4. Total gene diversity (Hy) ranges 6om 0.130 inWwttaZ/uMtAwj'yZonduMw.; to

15 0.242 in The low average values for Hg, ranging hrom 0.045 in #. to 0.079 in M demonstrated that little of the genetic variability within each species is present within individual populations. The relatively high average values for

G st, varying from 0.420 in#, canadensis to 0.688 in N. texanus, indicated that considerable genetic differentiation exists among the populations of each species; these valueswere higher than the averagesreported for annuals (Gst == 0.357), species with regional distributions (GgT= 0.216), and species of temperate regions (GsT= 0.246); they were comparable to those reported for selling species (Gst= 0.510) (Hamrick et al.

1990). Estimates of gene flow (Nm = (1 - Gsy) / 4Gsr) ranged from 0.113 in

texanM.; to 0.345 in #. Nm estimates of less than 1.0 suggest relatively little gene flow among populations (Slatkin et al. 1989).

The observed and expected (for randomly outcrossing species) frequencies of heterozygous loci in the three species of Nuttallanthus are presented in Table 5.

Observed heterozygosity was lower than expected heterozygosity for all polymorphic loci; the ratios of observed to expected heterozygosities were consistent with a selling breeding system in all three species.

Wright’s F statistics (Fis, Fn- and F st) for polymorphic loci within each species of

Nuttallanthus are provided in Table 6. The average amount of genetic variation distributed among populations (Fgy) ranged &om 0.430 in canndeMszs to

0.694 in N. texanus. For all three species, the average total inbreeding coefficient (Fit) appeared to be more greatly influenced by non-random mating within populations (Fis) than by differences in allele frequencies among populations (Fst)- Of the 144 single­ locus fixation indices tested, 138 (96 %) were both positive and significantly different

16 from Hardy-Weinberg expectations (p < 0.05), indicating a deficiency of heterozygotes at polymorphic loci in these species; an excess of heterozygotes was observed at a single locus (SOD) in a population of but this excess was not significant (Appendix 4).

Dendrograms summarizing the genetic similarities among populations and species

of were produced using the Distance Wagner procedure (Farris 1972) in

combination with Roger’s distance as modified by Wright (1978) and Cavalli-Sfbrza and

Edwards (1967) chord distance coefficients, as well as with UPGMA (Sneath et al. 1973)

in combination with Nei’s (1972) genetic identity and distance. All of the dendrograms

produced using the cited distance coefficients and clustering procedures were similar.

Nei’s (1972) genetic identity value resulted in the highest cophenetic correlation of the

dendrograms produced (0.920); consequently, that dendrogram is reproduced here

(Figure 2). The dendrogram clearly illustrates both geographic and taxonomic coherence;

conspecific populations cluster together and, within individual species, populations tend

to cluster with others from the same geographic region.Two discrete clusters of

populations of Nuttallanthus texanus are apparent; one cluster is predominately

composed of populations from the southeastern United States, California, Oklahoma and

central Texas, whereas populations from Arkansas, Louisiana, eastern Texas and

Oklahoma comprise the second cluster.

Nei’s (1972) genetic identities among populations of Nuttallanthus species were

subjected to nonmetric-multidimensional scaling (MDS) analysis; genetic similarity

among populations was plotted in three dimensions (Figure 3). Final stress (a measure of

the goodness of fit between the pairwise similarity coefficients in the original identity

17 matrix and those in the diagram produced by MDS) is 0.122, or “good to fair” (Kruskal

1964). The plot illustrates the pronounced genetic differentiation among the three species of jVwim/faMrAwf and distinguishes the two principal groups of M

Spearman’s rank-order correlation analysis was performed with SPSS for

Windows on matrices of Nei’s (1972) genetic distances and of linear distances (in kilometers) between populations. These analyses indicated positive and statistically- significant associations between genetic distance and linear distance between populations for each species: WwiinZZaMiAwj' (r^ = 0.382, p < 0.001), W.yZoncfanzty (r^ =

0.635, p < 0.001) and M reroMWj' (r^ = 0.380, p < 0.001).

CroMMg The results of seed production in three species of WnitaZ/aniAnj' following controlled pollinations are presented in Table 7. Individuals of the three species of Nuttallanthus examined in this study were autogamous and commonly produced cleistogamous flowers both early and late in the life cycle; in these flowers, the limbs of the undeveloped corolla securely enclosed the male and female reproductive organs and self-fertilization occurred in the flower bud. The capsules produced by cleistogamous flowers were smaller and produced significantly fewer seeds than did those of chasmogamous flowers. Pollen appeared to be required for successful fruit and seed development, because no emasculated, unpollinated flower produced mature capsules and seeds. Within each of the three species oî Nuttallanthus, no significant difference in seed production was observed between individuals pollinated with self pollen (geitonogamous hand pollination) or intraspecific cross pollen. Capsules produced through artificial hand pollination did yield significantly fewer seeds than did those from unmanipulated cleistogamous or chasmogamousflowers, which may have resulted from

18 limited pollen transfer between hand-pollinated flowers. The species oî Nuttallanthus examined in this study appeared to be cross-incompatible. Of the 240 hand pollinations representing all possible directional interspecific crosses, none resulted in the successful development of capsules and seeds. While individuals of the two principal groups of W texanus were readily distinguishable isozymically and morphologically, they appeared to be cross-compatible; no significant difference in seed production per fruit or seed germination rate was observed between the 35 hand pollinations made among individuals of the two principal groups ofN. texanus and the 45 hand pollinations made among individuals within each group.

No statistically significant difference in seed production within any treatment category was observed between individuals of j and W

Individuals of N. floridanus produced significantly fewer seeds per capsule for all treatments (other than emasculated, unpollinated flowers) compared to its congeners. No significant difference in the germination rate of seeds was observed among any species or treatment category.

DISCUSSION

texanus and N. floridanus exhibited moderate levels of genetic variation, comparable to that reported for other species with similar (regional to widespread) geographic ranges

(Hamrick 1989; Hamrick et al 1990). Much of the genetic variance present in plant species has been interpreted in terms o f population size and geographic range: species with small populations and restricted or narrowly-endemic distributions often possess

19 considerably less genetic diversity than do species with large population sizes and widespread ranges. Geographic range is the single best predictor of species-level genetic diversity (Hamrick et al 1990). This pattern is evident in where the more narrowly distributed M exhibited lower levels of genetic variation than did its widespread congeners (Table 3).

The level of genetic variation found within individual populations of the three

species was considerably lower than that observed at the species level.

Previous investigations and reviews have indicated that species-level genetic diversity is significantly and positively associated with population-level genetic diversity in plants, and that little (approximately 22%) of the genetic variation in a species occurs among populations of that species (Gottlieb 1981a; Hamrick et al. 1990). However,species- level and population-level genetic diversity are not strongly associated in species of

North American Although the percentage of polymorphic loci observed in

species (ranging from 64.3% to 71.4%) was well above the average reported for plant species (50.5%), most populations of NwhaZZuM/Am were hxed for a single allele at a majority of loci (Table 3; Appendix 1). In addition, Nuttallanthus species had from 42% (M ca»aù(eM.yû^) to 69% (A1 iexanwj:) of their genetic variation distributed among populations, whereas relatively little diversity existed within populations of these species (Table 4; Table 6). A number of life history characteristics have been shown to have profound influence upon the extent and organization of genetic variation within plant species (Gottlieb 1981a; Crawford 1990). Low chromosome numbers (n = 6 in species of Nuttallanthus; Crawford 2003) often are associated with low levels of infraspecific genetic variation (Hamrick 1989), and the maintenance of high

20 levels of genetic differentiation among populations has been associated with physical distance between populations (Wright 1969; Gibson et al. 1991); correlation analyses indicated positive and statistically-significant associations between genetic distance and the linear distance between conspecific populations oî Nuttallanthus. In addition, demographic, geographical (spacial substructuring within and among populations) and reproductive characteristics appear to affect the type and amount of genetic variation observed among North American species of Nuttallanthus and the development and maintenance of genetic differentiation among populations of these species.

A. and A. are herbaceous annuals

(occasionally biennials) that occupy serai and disturbed habitats and that produce large numbers of seeds per fruit and per individual (have high fecundity). Dramatic annual

changes in population size are observed often in these species (pers. observ.). In

addition, field observations and the early successional status of these adventive species

suggest that populations may be founded by a limited number of individuals. The

combination of founding effects and the occasional reduction of populations to a small

number of individuals could act to depress genetic variation within populations and to

promote differentiation among populations of these taxa. Population (sample) size was

significantly correlated with several commonly cited measures of genetic diversity in

species of Nuttallanthus, including the percentage of polymorphic loci, mean number of

alleles per locus, and the observed and expected levels of heterozygosity. These results

indicated that allozyme variation within these taxa is affected by population size and

suggest that the amoimt and type of genetic variation in these species has been afiected

by genetic drift (Nei et al. 1975).

21 The high genetic variability values observed in a small population ofN. texanus

(population 59), in comparison to the low variability estimates of other comparably-sized populations of this species (Table 3), as well as the notably-high level of populational genetic differentiation inM in comparison to its congeners (Table 4) suggest that spacial structure within and among populations may also influence populational genetic diversity in this species. The m^ority of the populations sampled in this study appeared to be structurally-homogeneous; however, population 59 consisted of two well-defined subpopulations: small groups of individuals were found growing in 2 “islands” of soil in shallow depressions in a granitic outcrop. While these subpopulations were separated by a distance of no more than 25 meters, individuals from each subpopulation were fixed for different alleles at 7 of 14 polymorphic loci (Appendix 2), dramatically increasing the estimates of genetic diversity in this population. The high level of genetic differentiation observed among populations of A. texawzty relative to its congeners may be attributed to the presence in this species of two largely allopatric and genetically-distinctive allelic groups of populations: one group (Group 1) composed of populations fromthe southeastern United States, California, Oklahoma and central Texas, and a second (Group

2) comprised of populations from Arkansas, Louisiana, eastern Texas and Oklahoma.

While these groups were fixed for the same allele at 4 of 14 polymorphic loci, quantitative allelic differences were observed at several loci (AAT-2, GPD, PGM-1, TPI-

2), at which populations of the two groups exhibited different fixed or high-frequency alleles; nine of the 31 alleles observed in Group 1 were not found in Group 2.

Other life history characteristics found in Nuttallanthus species associated with the promotion of populational genetic differentiation include traits that act to inhibit gene

22 flow among individuals and populations, such as seeds with limited dispersal capacity and an autogamous mating system. Seeds of Nuttallanthus are very small (approximately

300 microns in length) and are lacking in any suface ornamentation commonly associated with wind dispersal; these seeds appear to be gravity-dispersed. Indirect estimates of gene flow Table 4) in species of based on allozyme data indicate limited gene flow among populations of these taxa (Wright 1951; Slatkin et al. 1989); these estimates are comparable to those reported for selfmg species with small, gravity-

dispersed seeds (Hamrick 1989). Selfmg species generally possess lower absolute levels

of genetic variation than do species with mixed or outcrossing mating systems, and a

greater proportion of the variation within selfmg species is apportioned among individual

populations; conversely, outcrossing taxa tend to sequester higherlevels of genetic

variation within - rather than among - populations (Brown 1979; Loveless et al 1984).

The extent and apportionment of genetic variation observed in North American

was congruent with the predominately-selfing mating system of these taxa.

Individuals of these species commonly produce cleistogamous (obligately-selfmg)

flowers; the anthers of chasmogamous flowers generally dehisced prior to the opening of

the perianth, coating the immediately-adjacent and receptive stigmas with self^pollen and

potentially inhibiting outcrossing. No significant difference in seed production was

observed between individuals artificially pollinated with self pollen and those pollinated

with intraspecific cross pollen, indicating that these species are completely self-

compatible (Table 7). Field tests conducted in two Oklahoma populations of N. texanus

(and involving five treatments: open pollination, pollinator exclusion, floral emasculation

and pollinator exclusion, emasculation and open-pollination, and hand/self-pollination of

23 emasculated flowers) provided similar results. These tests indicated N. texanus is facultatively xenogamous; while many potential Lepidopteran and Hemipteran pollinators were observed visiting successive flowers among different plants, successful pollen transfer and seed set was rare (Crawford, unpublished data). The high level of selfmg observed in these species was consistent with the deficiency of heterozygotes observed at all polymorphic loci (Table 5; Appendix 4).

The observed patterns of allozyme divergence in North

American Nuttallanthus are consistent with recognition of three distinct species with no

infraspecific taxa: N. canadensis, N. floridanus and N. texanus, as proposed by Sutton

(1988). Analyses of genetic divergence readily distinguished three principal groups in the genus. Qualitative and quantitative allelic differences among species were evident at

many loci; 41% of the observed alleles were unique to one of the three species and 4

marker alleles were either fixed or present in high frequencies in all populations of a

single species (two in A! and one each in A. and A. feronwj^).

Pairwise genetic identity values between species (average of 0.580) were much lower

than infraspecific identities (average I value of 0.888; Table 2). Although a wide range

of infrageneric identity values hasbeen documented in plant species, the observed

interspecific genetic identity (I) values in were lower than those reported

for most congeneric angiosperm species (Crawford 1983) and the mean value of 0.67

reported by Gottlieb (1981a). These results indicate that considerable allozyme

divergence has occurred subsequent to the interruption of gene flow among these species.

The extent of genetic divergence among species is concordant with the observed

cross-incompatibility among species; no artificial interspecific cross pollinations resulted

24 in the development of mature capsules and seeds and no individuals of purported hybrid origin were observed in natural populations. Although cross-compatibilities vary within populations of some plants, our observations of interspecific cross-incompatibility in

supports hypotheses indicating a long period of isolation since spéciation

(Grant 1981; Raven 1977). Species of appear to have similar habitat requirements, are sympatric in a portion of their ranges (often occurring in dense, mixed populations of two or three species) and exhibit similar flowering phenologies; many potential Lepidopteran and Hemipteran pollinators were observed visiting successive

flowers among plants of different species (pers. observ.). While cleistogamy and the

early dehiscence of the anthers of chasmogamous flowers may act to limit interspecific

cross-pollination, no other premating isolating mechanism was evident among these

species. However, crossing studies indicated the presence of some pre- or post-zygotic

barrier to hybridization, and no evidence of hybridization between species was apparent

in the isozyme data.

The dendrograms summarizing the genetic similarities among populations of

Nuttallanthus were robust with respect to different distance measures and clustering

algorithms; all of the dendrograms delineated three principal groups of populations

corresponding to the three currently-recognized North American species of

(Figure 2). Similar results were obtained through nonmetric-multidimensional scaling

analysis (MDS) of genetic identities among populations (Figure 3). While these results

illustrated the considerable allozymic divergence among these species, they did not

clearly resolve phylogenetic relationships among species of These species

did not appear to differ in isozyme number, and the broadly similar estimates of genetic

25 divergence among species renders inferences regarding their evolutionary relationships suspect. Recently derived plant species commonly exhibit a subset of the genetic variation present in the progenitor species, and little or no qualitative divergence at isozyme loci is evident between progenitor-derivative species pairs (Crawford 1990;

Loveless et al. 1988). The extent o f the genetic divergence among species of

(low identity values and significant frequency differences among shared alleles) and the nature of that divergence (numerous unique and marker alleles in each species) provide no evidence of a recent derivation of these taxa.

While plants possessing cleistogamous flowers have been accorded inffaspecific status as L. canadensis forma cleistogama Fernald (Femald 1936), field observations and breeding system studies indicate that many (if not all) individuals of the three species of

Nuttallanthus are capable of producing cleistogamousflowers both early and late in the life cycle; consequently, the recognition of infraspecific cleistogamous taxa in this genus is unjustified.

Infi-aspecific identity values in W (I = 0.819) were considerably lower than those found among populations of its congeners (I = 0.936 in W. canadensis and I =

0.909 in N. floridanus), demonstrating that considerable interpopulational genetic divergence has occurred in this species. Identity values within the two principal groups of W (I = 0.878 in Group 1 and I = 0.928 in Group 2) were comparable to those of other Nuttallanthus species, and the degree of genetic similarity between Groups 1 and

2 (I = 0.737) was intermediate between the average infraspecific (I = 0.888) and infirageneric (I = 0.580) values observed in Populations of the two groups of N. texamAy were largely allopatric (with the exception of population 58), a situation

26 observed in other infraspecific taxa exhibiting lowered genetic identities (Crawford

1989). Nine of the 31 alleles observed in Group 1 were absent in Group 2, which possessed no unique alleles; Group 2 contained a subset of the total genetic variation present in Group 1 (Appendix 2). These results suggest that gene exchange has been

restricted between these groups of populations for a considerable period of time.

Whereas allozyme data indicated a substantial degree of genetic divergence between

these two groups, breeding system experiments illustrated no reduction in interfertility.

No significant difference in the production or germination rate of seeds was observed

between the artificial pollinations made between individuals of Groups 1 and 2 and the

pollinations made among individuals within each group; however, the study provided no

information regarding the post-germination viability and fertility of the hybrid offspring,

so the potential for hybrid sterility or hybrid (Fz) breakdown may exist.

The extant center of taxonomic diversity of

Nuttallanthus lies in the east Gulf Coastal Plain of Alabama and Florida where all three

species are sympatric. This region harbors a large number of endemic species and

genera, has been recognized as a center of plant endemism, and may have served as a

refuge for many plant species during cycles of Pleistocene glaciation (Delcourt et al.

1981; Estill et al. 2001). While the flora of the Atlantic and Gulf coastal plain has been

considered to be of geologically-recent origin (Thome 1993), the large number of genera

endemic to this region indicates that some floristic elements are of greater antiquity

(Sorrie et al. 2001). Nei's (1987) stepwise mutation rate models (ÏE and IgA) based on

genetic identities among species of Nuttallanthus provided estimates of the elapsed time

since divergence of A. and A. ranging from 2.75 to 4.0 million years

27 (my) BP; estimated divergence times for and jV. ranged 6om 3.0 to

4.5 my BP, and those for N. canadensis and N. floridanus ranged from 5.0 to 7.0 my BP.

These models suggested that species ofNuttallanthus diverged in the late Tertiary (late

Miocene and Pliocene). Although much of the Atlantic and Gulf coastal plain was inundated by advancing seas during portions of the Tertiary and Pleistocene, isolated areas of the coastal plain were suitable for plant habitation by the late Miocene (Walker et al. 1987). These 6agmented areas of the coastal plain apparently served as réfugia for a number of archaic plant taxa (Sorrie et al. 2001) and may have provided opportunities for spéciation among geographically-isolated ancestral populations of Nuttallanthus.

In addition to the extensive genetic divergence between species of JVwtfal/aMtAw.;, individual taxa exhibited genetic substructuring at a regional level. Significant correlations were evident between genetic distance and the physical distance between conspecific populations of Nuttallanthus, and clustering and ordination analyses consistently grouped conspecific populations with others from the same geographic region. Many of these population clusters are located in geographic regions corresponding to recognized centers of plant endemism, where environmental conditions have fostered spéciation or allowed réfugiai taxa to persist. UPGMA of Nei's (1972) genetic identities among populations (Figure 2) illustrated genetic divergence between three groups of A.canadensis: 1) populations located on the Delmarva peninsula and separated from southern and western conspecific populations by the Chesapeake Bay and the Potomac River; 2) populations found on the Mid-Atlantic Coastal Plain of Virginia,

North Carolina and Georgia and the Outer Banks of North Carolina; and 3) populations

located on the Southern Atlantic and Gulf coasts of the United States (and one population

28 from Massachusetts). The modem Chesapeake Bay corresponds to the geologic

Salisbury embayment, which has been recognized as a major biogeographic barrier; many plant species found in the middle- and southern Atlantic and Gulf coastal plain do not occur to the north of this break (Sorrie et al. 2001). The Mid-Atlantic Coastal Plain and the Apalachicola region in the Panhandle of Florida are separated by several large river

systems (including the Cape Fear and Savannah Rivers) which have influenced the distributional patterns of many plant taxa (Estill et al. 2001) and which have apparently

formed partial barriers to gene flow among populations of #. The clustering

of the Massachusetts population (20) with those of the Southern Atlantic and Gulf coast populations conforms to no obvious biogeographicpattern; the location of population 20

may be due to anthropogenic dispersal of seeds. N. canadensis and N. texanus commonly

occur in agricultural settings, including fallow fields and stands of alfalfa

the seeds of AwttaZZantAwj are similar in size to those of alfalfa and may function

as crop mimics. However, the pronounced spatial and genetic structuring of populations

within Nuttallanthus suggests that such anthropogenicdispersal has been relatively rare.

Populations of N. floridanus also exhibited a pronounced geographic pattern in

divergence at isozyme loci, with two groups of populations occurring on the Gulf coastal

plain and separated by the Apalachicola River, and a single outlier located in central

pensinsular Florida, The Apalachicola region of the Florida Panhandle and central

peninsular Florida have been identified as centers of plant endemism, and the

Apalachicola/Chattahoochee River system has constituted a barrier to migration and gene

flow in a number of coastal plain taxa (Avise 1992; Avise et al. 1979; Sorrie et al. 2001).

29 Concordance between geographic isolation and genetic differentiation among populations is also apparent in N. texanus-, UPGMA identified clusters of populations located in California, Oklahoma and Texas, and in the southeastern United States (Figure

2). The populations comprising group 2 of this species are largely allopatric from those of group 1 and are located in and near the former Mississippi Embayment.

30 LITERATURE CITED

AVISE, J. C. 1992. Molecular population structure and the biogeographic history of a regional fauna: a case history with lessons for conservation biology. Oikos 63: 62-76.

AVISE, J. C., C. GIBLIN-DAVIDSON, J. LAERM, J. C. PATTON AND R. A.

LANSMAN. 1979. Mitochondrial DNA clones and matriarchal phylogeny within and

among geographicpopulations of the pocket gopher, Geomys pinetis. Proc. Natl. Acad.

Sci. 76: 6694-6698.

BERG, E. E. AND J. L. HAMRICK. 1997. Quantification of genetic diversity at

allozyme loci. Canadian Journal of Forest Resouces 27: 415-424.

BIGAZZI, M. 1989. Occurrence, ultrastructure and developmental features of nuclear

inclusions in the Tribe Antirrhineae (Scrophulariaceae). I. Amorphous inclusion.

Carylogia 42 (3-4): 313-328.

BIGAZZI, M. 1989. Occurrence, ultrastructure and developmental features of nuclear

inclusions in the Tribe Antirrhineae (Scrophulariaceae). II. Tubular inclusion. Carylogia

42 (3-4): 329-343.

BIGAZZI, M. 1993. A survey on the intraclear inclusions in the Schrophulareaceae and

their systematic significance. Nord. J. Bot. 13: 19-31.

BROWN, A. H. D. 1979. Enzyme polymorphism in plant populations. Theoretical

Population Biology 15: 1-42.

CAVALLI-SFORZA, L. L. AND A. W. F. EDWARDS. 1967. Phylogenetic analysis:

models and estimation procedures. Evolution 21: 550-570.

CHAPMAN, A. W. 1860. Flora of the Southern United States: 290-291. Iverson, Phinney

and Co. New York.

31 CRAWFORD, D. J. 1983. Phylogenetic and systematic inferences from electrophoretic studies, /n S. D. TANKSLEY AND T. J. ORTON. Isozymes in plant genetics and breeding, part A: 257-287. Elsevier Science Publishers, Inc., New York.

CRAWFORD, D. J. 1989. Enzyme electrophoresis and plant systematics. 7n D. SOLTIS

AND P. SOLTIS [eds.], Isozymes in Plant Biology: 146-164. Advances in Plant Sciences

Series, Vol. 4. Dioscorides Press, Portland.

CRAWFORD, D. J. 1990. Plant Molecular Systematics. John Wiley and Sons, New

York.

CRAWFORD, P.T. 2003. Biosystematics of North American species of AwW/antAwf

(Lamiales). Ph.D. dissertation. University of Oklahoma: Norman, OK.

CRONQUIST, A., A.H. HOLMGREN, N.H. HOLMGREN, J.L. REVEAL, AND P.K.

HOLMGREN. 1984. Intermountain flora: Vascular plants of the intermountain west,

U.S.A. Vol. 4, The New York Botanical Garden, Bronx, New York.

DELCOURT, P. A. AND H. R. DELCOURT. 1981. Vegetation maps for eastern North

America: 40,000yr B.P. to the present. In R. C. Romans [ed.] Geobotany II: 123-165.

Plenum Press, New York.

DUMONT DE COURSET, GEORGES LOUIS MARIE. 1802. Bot. Cult. 2: 96.

ELISENS, W. J. AND A. S. TOMB. 1983. Seed morphology in New World Antirrhineae

(Scrophulariaceae): systematic and phylogenetic implications. PI. Syst. Evol. 142: 23-47.

ELISENS, W. J. 1985. The systematic significance of seed coat anatomy among new world species of Tribe Antirrhineae (Scrophulariaceae). Systematic Botany 10(3): 282-

299.

32 ELISENS, W. J. 1986. Pollen morphology and systematic relationships among new world species in Tribe Antirrhineae (Scrophulariaceae). Amer. J. Bot. 173(9): 1298-1311.

ELISENS, W. J. AND D. J. CRAWFORD. 1988. Genetic variation and diflerentiation in the genus Mabrya (Scrophulariaceae-Antirrhineae): systematic and evolutionary

inferences. American Journal of Botany 75(1): 85-96.

ESTILL, J. C. AND M. B. CRUZAN. 2001. Phytogeography of rare plant species

endemic to the southeastern United States. Castanea 66 (1-2): 3-23.

FARRIS, J. S. 1972. Estimating phylogenetic trees from distance matrices. American

Naturalist 106: 645-668.

FERNALD, M. L. 1936. Plants from the outer coastal plain of Virginia (concluded).

Rhodora 38: 414-452.

FERNALD, M. L. 1943. Virginia Botanizing under Restrictions. Rhodora 45: 357-413,

445-480, 485-511.

GIBSON, J. P. AND J. L. HAMRICK. 1991. Genetic diversity and structure in fm it;

(table mountain pine) populations. Can. J. For. Res. 21: 635-642.

GOTTLIEB, L. D. 1981a. Electrophoretic evidence and plant populations. In L.

REIN HOLD, J. B. HARBORNE AND T. SWAIN [eds.], Progress in phytochemistry,

Vol. 7: 1-46. Pergamon, New York.

GOTTLIEB, L. D. 1981b. Gene number in species of Astereae that have different

chromosome numbers. Proceedings of the National Academy of Sciences U.S.A. 78:

3726-3729.

GOTTLIEB, L. D. 1982. Conservation and duplication of isozymes in plants. Science

216: 373-380.

33 GRANT, V. 1981. Plant Spéciation, 2""^ edition. Columbia University Press, New York.

HAMRICK, J. L. 1987. Gene flow and the distribution of genetic variation in plant populations. 7» K. URBANSKA [ed.], Proc. lOPB Sym. Differentiation in Higher Plants:

53-67. Academic Press, New York.

HAMRICK, J. L. 1989. Isozymes and the analysis of genetic structure in plant populations. 7n D. SOLTIS AND P. SOLTIS [eds.]. Isozymes in Plant Biology: 87-105.

Dioscorides Press, Washington, DC.

HAMRICK, J. L. AND M. J. W. GODT. 1990. Allozyme diversity in plant species. A.

H. D. BROWN, M. T. CLEGG, A. L. KAHLER AND B. S. WIER [eds.]. Plant population genetics, breeding and genetic resources: 43-63. Sinauer Associates, Inc.

Sunderland, Massachusetts.

HITCHCOCK, L. AND A. CRONQUIST. 1998. Flora of the Pacific Northwest, 11th ed.

University of Washington Press, Seattle, WA.

KRUSKAL, J. B. 1964. Nonmetric multidimensional scaling: a numerical method.

Psychometrica 29: 28-42.

LINNAEUS, C. 1753. Species Plantarum. 2 vols. Stockholm.

LOVELESS, MD AND JL HAMRICK. 1988. Genetic organization and evolutionary history in two North American species of Cirsium. Evolution 42: 254-265.

LOVELESS, MD AND JL HAMRICK. 1984. Ecological determinants of genetic structure in plant populations. Ann. Rev. Ecol. Syst. 15: 65-95.

MILLER, P. 1754. The Gardener’s Dictionary abridged ed. 4, 2. London.

MILLER, P. 1768. The Gardener’s Dictionary ed. 8. London.

34 MUNZ, P. A. 1926. The Antirrhinoideae-Antirrhineae of the New World. Proc. Calif.

Acad. Sci. 15: 323-397.

NEI, M. 1972. Genetic distance between populations. American Naturalist 106:283-292.

NET, M. 1973. Analysis of gene diversity in subdivided populations. Proceedings of the

National Academy of Sciences U.S.A. 70: 3321-3323.

NEI, M. 1977. F-statistics and analysis of gene diversity in subdivided populations. Ann.

Human Genet. 41: 225-233.

NEI, M, 1987. Molecular Evolutionary Genetics. Columbia University Press. New York.

NEI, M., T. MARUYAMA AND R. CHAKRABORTY. 1975. The bottleneck effect and genetic variability in populations. Evolution 29: 1-10.

PENNELL, F. W. 1920. Scrophulariaceae of the southeastern United States. Contr. New

York Botanical Garden 221: 224-291, reprinted 6om Proc. Acad. Nat. Sci. Philadelphia,

1919.

PENNELL, F. W. 1922. Proc. Acad. Nat. Sci. Philadelphia 73:502.

PENNELL, F. W. 1935. The Scrophulariaceae of eastern temperate North America.

Academy of Natural Sciences of Philadelphia, Monograph No. 1: i-xv, 1-650.

RAVEN, P. H. 1980. Hybridization and the nature of species in higher plants. Canadian

Botanical Association Bulletin (supplement) 13: 3-10.

ROHLF, F. J. 2002. NTSYSpc: Numerical Taxonomy System, ver. 2.11c. Exeter

Publishing, Ltd.: Setauket, NY.

ROTHMALER, W. 1954. Notes on western Antirrhineae. Leafl. West. Bot. 7: 113-117.

SCHEELE, A. 1848. Reitrage sur flor von Texas. Linnaea 21: 747-768.

35 SLATKIN, M. AND N. H. BARTON. 1989. A comparison of three indirect methods for estimating average levels of gene flow. Evolution 43: 1349-1368.

SMALL, E., S. I. WARWICK AND B. BROOKES. 1999. Allozyme variation in relation to morphology and taxonomy in Medicago sect. Spirocarpos subsect. Intertextae

(Fabaceae). Plant Systematics and Evolution 214: 29-47.

SNEATH, P. H. AND R R. SOKAL. 1973. Numerical Taxonomy. WH Freeman, San

Fransisco.

SOLTIS, D. E., C. H. HAUFLER, D. C. DARROW AND G. J. GASTONY. 1983. Starch gel electrophoresis of ferns: a compilation of grinding buffers, gel and electrode buffers, and staining schedules. American Fern Journal 73: 9-27.

SORRIE, B. A. AND A. S. WEAKLEY. 2001. Coastal plain vascular plant endemics: phytogeographic patterns. Castanea 66 (1-2): 50-82.

SUTTON, D.A. 1988. A revision of the tribe Anthirrhmeae. Oxford Univ. Press, London.

SWOFFORD, D. L. AND R. B. SELANDER. 1981. BIOSYS-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. Journal of Heredity 72: 281-283.

THORNE, R. F. 1993. Phytogeography.In Flora of North America Editorial Committee

[eds.] Flora of North America North of Mexico, Volume 1, Introduction: 132-153.

Oxford University Press, New York.

TUTIN, T. G., V. H. HEYWOOD, N. A. BURGES, D. M. MOORE, D. H.

VALENTINE, S. M. WALTERS AND D. A. WEBB [eds.] 1972. Flora Europaea

Volume 3, Diapensiaceae to Myoporaceae. Cambridge University Press, London.

36 USDA, NRCS. 2002. The PLANTS Database, Version 3.5 (http://plants.usda.gov).

National Plant Data Center, Baton Rouge, LA 70874-4490. U.S.A.

WALKER, H. J. AND J. M. COLEMAN. 1987. Atlantic and Gulf Coastal Province. /»

GRAF, W. L. [ed.] Geomorpbic Systems of North America. Geological Society of

America, Centennial Special Volume 2: 51-110. Boulder, CO.

WEEDEN, N. F. AND J. F. WENDEL. 1989. Genetics of plant isozymes. 7» D. E. Soltis and P. S. Soltis [eds.], Isozymes in plant biology, 46-72. Dioscorides, Portland, OR.

WHITKUS, R. 1988. Modified version of GENESTAT: a program for computing genetic statistics from allele frequency. Plant Genetics Newsletter 4:10.

WRIGHT, S. 1951. The genetic structure of populations. Annals of Eugenics 15: 323-

354.

WRIGHT, S. 1965. The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution 19: 395-420.

WRIGHT, S. 1969. Evolution and the genetics of populations, Vol 2. University of

Chicago Press, Chicago, IL.

WRIGHT, S. 1978. Evolution and the genetics of populations, Vol 4. Variability within and among natural populations. University of Chicago Press, Chicago, IL.

37 Table 1. Numbered collection data for 60 populations representing three species of MrttoZ/antAwA

examined for isozyme variation. Population codes are given in bold numbers; OTUs included in

population level isozyme analyses are indicated by an asterisk (*). Numbers of individuals per

OTU examined for isozyme variation are given in parentheses, followed by the collection

numbers for those specimens.

Alabama: Baldwin Co.: 1* = 30.414N 87.598W, fallow cornfield 0.1 mi. NE of Highway 98 & S

Michigan, Elberta, AL (n=16, Crawford 358-373); 2 = 30.238N 87.882W, open pine woods N of

Miller Memorial Cemetery, Miller Memorial Road and Highway 180 intersection (n=5,

686-690). Mobile Co.: 3 = 30.433N 88.144W, grounds of Bellingrath Gardens Estate, near picnic

grounds (n=10, Crawford 324-333); = 30.243N4 88.078, E end of Dauphin island in seaside

dunes (n=2, Crawford dJO-dJ/j.

Delaware: Sussex Co.: 5* = 38.574N 75.056W, sand dunes in meadow W of Highway 1, 2.0 mi.

N of Highway 1 & Highway 26 intersection (n=30, Cmn^rd 2^7-240^.

Florida: Bay Co.: 6* = 30.204N 85.847W, S of Beach Front Road, 3.0 mi. W of Highway Alt98

& Highway 392 intersection (n=25, Crovt^rd 3&0-3P9, 774-77^; Calhoun Co.: 7 = 30.464N

85.045W, fallow cornfield 1.0 mi. N of Blountstown (n=8, Crawford 853-860). Franklin Co.: 8*

= 29.833N 84.876W, meadow S of Highway 65, 8.6 mi. N of Highway 65 & Highway 98

intersection (n==19, Cran^rd 40d-475, 477-42.^; 9* == 29.909N 84.394W, dunes on Lighthouse

Point, W of Alligator Point village, W end of peninsula (n=13, Cmn^rd 449-4d7j; 10 = 29.853N

84.664W, open pine woodland near Carrabelle, N of 3’"'^ Street (n=3, Crawford 808-810).

Lafayette Co.: 11* = 30.139N 83.290W, improved pasture 0.1 mi. N of Highway27, 8.8 mi. W

of Mayo (n=10, Cranÿ%)rd 493-5027. Wakulla Co.: 12* = 30.136N 84.326W, meadow 0.1

38 Table 1 cont.

mi. NE of Highway 98 & Spring Creek Highway intersection (n=10, Crawford 817-826). Walton

Co.: 13 = 30.309N 86.102W, shore of Eastern Lake, E of Seagrove Beach, near Highway 30A

(n=5,

Georgia: Camden Co.: 14* = 30.759N 81.658W, meadow above N bank of St. Mary's River, 0.1 mi. E of 195 (n=20, Crawford 509-528). Candler Co.: 15* = 32.355N 81.989W, meadow 0.3 mi.

N of Interstate 1 6, 11.7 mi. SE of Interstate 16 & Highway 57 intersection (n=10, Crawford 596-

605). Glynn Co.: 16* = 31.020N 81.435W, meadow near St. Andrew picnic area, S Riverside

Drive, Jekyll Island (n=10, Liberty Co.: 17* = 31.675N 81.414W, edge of

Beltowne marsh 3.6 mi. S of Retreat (n=10,

Maryland: Caroline Co.: 18* = 38.817N 75.748W, wet meadow S of Highway 404, 1.5 mi. E of

Highway 404 & Highway 16 intersection (n-10, Cranybrtf 244-2J3). Worcester Co.: 19* =

38.096N 75.499W, fallow cornfield E of Highway 113, 0.3 mi. S of mile 6 marker (n=17,

Massachusetts: Middlesex Co.: 20* = 42.S04N 71.265W, sandy roadside at edge of pine woodland, Concord Field Station, Bedford (n=15, Crow/brtf 907-975).

North Carolina: Currituck Co.: 21* = 36.278N 75.915W, 0.5 mi. N of Highway 158 & Highway

3E intersection (n-10, Cmw/brd 77&72;^. Dare Co.: 22* = 35.261N 75.579W, Hatteras Island, near intersection of Highway 12 & Paradise Lane (n=10, 704-773;. Duplin Co.: 23* -

34.926N 77.652W, S of Highway 24, 1 mi. W of Duplin/Onslow county line (n=10, Crawford

91-100). Hoke Co.: 24* = 35.007N 79.305W, SW of intersection of Highway 211 & SRI202

(n-20, 6&&T).

39 Table 1 cont.

South Carolina: Beaufort Co.: 25* = 32.377N 80.832W, meadow on W bank of Chechessee R,

0.1 mi. SE of Highway 170 bridge (n=10, Crawford 573-582).

Virginia: Accomack Co.: 26* = 37.912N 75.356W, Chincoteague Island, S of Highway 175,

near causeway to Assateague Island (n=20, 760-779^. Northampton Co.: 27* =

37.145N 75.967W, meadow 0.2 mi. E of Highway 13, 0.4 mi. S of Highway 13 & Latimer Siding

Road intersection (n=20, Cmw/ôr(7 73'7-7jjj. Orange Co.: 28* = 38.261N 77.980W, near

Highway 20 & Village Road intersection (n:=10, Crow/or(7 257-26^.

Alabama: Baldwin Co.: 29* = 30.238N 87.882W, open pine woods N of Miller Memorial

Cemetery, Miller Memorial Road and Highway 180 intersection (n=l 1, Craw/br<7 672-d&2).

Mobile Co.: 30* - 30.243N 88.078, E end of Dauphin island in seaside dunes (n=l 1, Crmv/br(7

6^7-667/

Florida: Bay Co.: 31* = 30.204N 85.847W, S of Beach Front Road, 3.0 mi. W of Highway

Alt98 & Highway 392 intersection (n=25, 723-74:;^. Franklin Co.: 32* = 29.909N

84.394W, dunes on Lighthouse Point, W of Alligator Point village, W end of peninsula, (n=10,

467-47^; 33* = 29.723N 84.890W, 0.2 mi. S of Highway 98 & Highway 300

intersection (n=8, Crau^r(7 733-7#; 34* = 29.724N 84.899W, sandy meadow SE of E end of

Gorrie Bridge, E bank of Apalachicola River (n^20, 772-797j; 35* = 29.853N

84.664W, open pine woodland near Carrabelle, N of 3"^ Street (n=10, Craw/ôr(7 7P7-gOg^.

Putnam Co.: 36* = 29.623N 81.912W, open pine woodland 0.1 mi. S of Highway 20, W of

Interlachen (n=15, Crauÿhrcl 337-337/

40 Table 1 cont.

Alabama: Baldwin Co.: 37* = 30.414N 87.598W, fallow cornfield 0.1 mi. NE of Highway 98 &

S Michigan, Elberta (n=14, 3 3 9 - Mobile Co.: 38* = 30.243N 88.078, E end of

Dauphin island in seaside dunes (n=10, Crawford 639-648).

Arkansas: Conway Co.: 39* = 35.171N 92.755W, ca. 0.1 mi. S of Interstate 40 mile marker 107

(n=10, Cron^rd 32-67/ Crawford Co.: 40* = 35.528N 94.041W, ca. 0.25 mi. S of Interstate 40,

1.0 mi. W of Mulberry exit (n-10, 39-4(^; Logan Co.: 41 = 35.300N 93.634W,

Subiaco Academy grounds, Subiaco (n=5, 270-27/) [Logan].

California: Monterrey Co.; 42* = 36.51 IN 12I.942W, grown from seed collected at Point Lobos

State Reserve (n=5, Cmu/bral 920-92/). Santa Barbara Co.: 43* = 34.044N 119.718W, grown from seed collected at Pelican Bay, Santa Cruz Island (n=5, 929-933/

Florida: Calhoun Co.: 45 = 30.464N 85.045W, fallow cornfield 1.0 mi. N of Blountstown (n=4,

Crmy/brcf g62-&6/). Franklin Co.: 46 = 29.833N 84.876W, 8.6 mi. N of Highway 65 & Highway

98 intersection (n-1, 476); 47* = 29.909N 84.394W, dunes on Lighthouse Point, W of

Alligator Point village, W end of peninsula (n=13, Crou/brcf 437-44/); 48 = 29.723N 84.890W,

0.2 mi. S of Highway 98 & Highway 300 intersection (n=5, Crau^r<7 762-76t^; 49 = 29.853N

84.664W, open pine woodland near Carrabelle, N of 3"^ Street (n=l, Crou^rcf &77/ 50* =

Wakulla Co.: 30.136N 84.326W, meadow 0.1 mi. NE of Highway 98 & Spring Creek Highway intersection (n-13, Craw/br(7 4gO-4g9, ,329-g37). Walton Co.: 51* = 30.309N 86.102W, shore of

Eastern Lake, E of Seagrove Beach (n=7, Crau/br(7 696-702).

Georgia: Candler Co.: 52* = 32.355N 81.989W, meadow 11.7 mi. SE of Interstate 16 and

Highway 57 intersection (n=5, 3<37-397).

41 Table 1 cont.

Louisiana: St. Landry Parish: 53* = 30.540N 92.028W, meadow 0.2 mi. E of Interstate 49 &

Highway 190 intersection (n=10, 296-30.^. St. Tammany Parish: 54* = 30.298N

89.666W, meadow 0.3 mi. S of Interstate 12, high on W bank of Pearl River (n=12, Crawford

Oklahoma: Cleveland Co.: 56* = 35.214N 97.328W, T9N RIW Sec 32 NW 1/4 (n=10,

CmnybfY/ 626-633/ Garvin Co.: 57* = 34.708N 97.543W, T3N R3W Sec 30 NE 1/4 (n=20,

606-62/); 58* = 34.745N 97.548W, T3N R3W Sec 19 SE 1/4 (n=25,

963/ Johnston Co.: 59* = 34.327N 96.770W, T3S R5E Sec 3 NE 1/4, "islands" of shallow soil in

Tishomingo granite of Ten-acre Rock (n=10, 333-39:^.

South Carolina: Beaufort Co.: 60* 32.377N 80.832W, meadow on W bank of Chechessee R,

0.1 mi. SE of Highway 170 bridge (n=10, Crawford 559-568).

Texas: Harrison Co.: 61* = 34.470N 94.595W, pasture 0.1 mi.N of Interstate 20, 0.8 mi. W of exit 604 (n=10, Crawford 283-292). Smith Co.: 62* = 32.469N 95.389W, improved pasture 0.25 mi. S of Interstate 20 near mile marker 557 (n=15, Crauÿbrcf 377-33/).

42 Table 2. Mean values for Nei's (1972) Genetic Identity (I) coeGIcients for pairwise comparisons of 50 populations within and among three species of Ranges are given in parentheses; numbers of sampled populations / individuals are given in brackets.

Species /V. caMadg/iyiy /V. A/. texuMMj'

0.936 [22/325] (0.834 - 0.998)

0.516 0.909 [8/110] (0.408 - 0.655) (0.817- 1.000)

A/ 0.623 0.601 0.819

[20/214] (0.449 - 0.765) (0.429 - 0.770) (0.604 - 1.000) Table 3. Summary of allozyme variation for 14 putative loci within 22 populations o f 8 populations of

Æ yZorw&znwf and 20 populations of TVi f&tüMwj. TV = sample size, W = mean number of alleles per locus, = mean number

of alleles per polymorphic locus, f = % polymorphic loci, Tif ^ = mean observed heterozygosity, /fg = mean expected

heterozygosity, and ^4, = effective number of alleles.

Population NA P Ho Ho Ao

1 16 1.50 2.17 42.9 0.004 0.162 1.19

5 30 1.29 2.00 28.6 0.010 0.101 1.11

6 25 1.50 2.00 50.0 0.006 0.119 1.14

8 19 1.43 2.00 42.9 0.011 0.098 1.11

9 13 1.43 2.20 35.7 0.000 0.140 1.16

11 10 1.14 2.00 14.3 0.007 0.045 1.05

12 10 1.07 2.00 7.1 0.000 0.034 1.04

14 20 1.29 2.00 28.6 0.011 0.055 1.06

15 10 1.29 2.00 28.6 0.000 0.069 1.07

16 10 1.14 2.00 14.3 0.021 0.067 1.07

17 10 1.14 2.00 14.3 0.000 0.064 1.07 Table 3 cont.

Population N AP A,

18 10 1.29 2.00 28.6 0.007 0.078 1.08

19 17 1.36 2.00 35.7 0.008 0.105 1.12

20 15 1.00 - 0.0 0.000 0.000 1.00

21 10 1.14 3.00 7.1 0.000 0.033 1.03

22 10 1.07 2.00 7.1 0.007 0.027 1.03

23 10 1.36 2.00 35.7 0.007 0.137 1.16 en 24 20 1.21 2.00 21.4 0.004 0.091 1.10

25 10 1.21 2.00 21.4 0.000 0.087 1.10

26 20 1.36 2.25 28.6 0.004 0.071 1.08

27 20 1.29 2.00 28.6 0.014 0.092 1.10

28 10 1.00 - 0.0 0.000 0.000 1.00 Mean 14.8 1.25 2.08 23.7 0.006 0.076 1.09 (SD) 5.74 0.147 0.224 13.9 0.005 0.042 0.049 Species level 325 2.14 2.60 71.4 0.006 0.134 1.15 (SD) 0.915 0.663 0.154 Table 3 cont.

Population N P ^0

29 11 1.00 - 0.0 0.000 0.000 1.00

30 11 1.00 - 0.0 0.000 0.000 1.00

31 25 1.21 2.00 21.4 0.000 0.089 1.10

32 10 1.29 2.00 28.6 0.000 0.079 1.09

33 8 1.29 2.33 21.4 0.000 0.069 1.07 4^ o\ 34 20 1.14 2.00 14.3 0.000 0.020 1.02

35 10 1.07 2.00 7.1 0.000 0.034 1.04

36 15 1.21 2.00 21.4 0.005 0.053 1.06 Mean 13.8 1.15 2.06 14.3 0.001 0.043 1.05 (SD) 5.52 0.111 0.123 10.1 0.002 0.033 0.036 Species level 110 1.86 2J3 64J 0.001 0.119 1.14 (SD) ------0.742 0.471 ------0.179 Table 3 cont.

Population NAP ^ 0

37 14 1.50 2.17 42.9 0.000 0.152 1.18

38 10 1.14 2.00 14.3 0.000 0.026 1.03

39 10 1.00 - 0.0 0.000 0.000 1.00

40 10 1.07 2.00 7.1 0.000 0.013 1.01

42 5 1.00 - 0.0 0.000 0.000 1.00

43 5 1.00 - 0.0 0.000 0.000 1.00

47 13 1.21 2.00 21.4 0.000 0.079 1.09

50 13 1.29 2.00 28.6 0.011 0.115 1.13

51 7 1.14 2.00 14.3 0.054 0.050 1.05

52 5 1.00 - 0.0 0.000 0.000 1.00

53 10 1.21 2.00 21.4 0.007 0.084 1.09

54 12 1.36 2.00 35.7 0.012 0.124 1.14

56 10 1.14 2.00 14.3 0.000 0.036 1.04

57 20 1.29 2.00 28.6 0.007 0.104 1.12 Table 3 conl.

Population NA A.

58a 13 1.57 2.14 50.0 0.005 0.194 1.24

58b 12 1.14 2.00 14.3 0.012 0.055 1.06

59 10 1.64 2.13 57.1 0.000 0.283 1.39

60 10 1.14 2.00 14.3 0.000 0.026 1.03

61 10 1.07 2.00 7.1 0.000 0.013 1.01

62 15 1.21 2.00 21.4 0.000 0.104 1.12 4^ oo Mean 10.7 1.21 2.03 19.6 0.005 0.073 1.09 (SD) 3.54 0.183 0.058 163 0.012 0.073 0.096

Species level 214 2.21 2.70 71.4 0.005 0.234 I J l (SD) 0.860 0.458 ------0.223 Table 4. Genetic diversity statistics for W.yZoricfamwf andÆ 77^ = total gene diversity, Tfg == gene diversity within populations, Z) gy ^ gene diversity among populations, = the proportion of gene diversity apportioned among populations, and - Wright's gene flow estimate = (1-Ggy) / 4G_^y].

Numbers o f sampled populations / individuals are given in brackets.

Species Hr ;^s DsT G ST

0.1361 0.0790 0.0572 0.4199 0.3454 [22 / 325] 4^ ‘O 0.1301 0.0449 0.0853 0.6553 0.1315 [8/110]

A/ iexa/nty 0.2422 0.0756 0.1666 0.6879 0.1134 [20/214] Table 5. Observed and expected 6equencies of heterozygotes across 14 polymorphic loci in three species of Nuttallanthus. n = number of sampled individuals.

A/ A/ yZonVfanwj' A/ texo/ma'

n = 325 n = 110 n = 214

Locus Obs. Exp. Obs. Exp. Obs. Exp.

AAT-1 0.0 0.0 0.0 0.0 0.0 0.0975

AAT-2 0.0154 0.3646 0.0 0.0180 0.0047 0.3102

ADH-1 0.0 0.0539 0.0 0.4628 0.0 0.0980

GPD 0.0 0.0713 0.0 0.0 0.0 0.4807

IDH-1 0.0154 0.2467 0.0091 0.2014 0.0 0.0545

MDH-1 0.0 0.0596 0.0 0.1040 0.0 0.4686

MDH-4 0.0 0.0 0.0 0.0701 0.0 0.0

6PGD-2 0.0 0.0 0.0 0.0180 0.0 0.0

PGI-1 0.0 0.0423 0.0 0.0 0.0 0.0

PGI-2 0.0277 0.2788 0.0 0.5902 0.0280 0.3272

PGM-1 0.0 0.0303 0.0 0.1653 0.0 0.4470

SOD 0.0062 0.3921 0.0 0.0 0.0 0.0

TPI-1 0.0 0.0 0.0 0.0 0.0187 0.3107

TPI-2 0.0215 0.4200 0.0 0.0874 0.0 0.6666

Mean 0.0062 0.1400 0.0007 0.1227 0.0037 0.2329

Ratio 0/E 0.0443 0.0057 0.0159

50 Table 6. Wright's (1978) F statistics averaged across all populations o f three species of n = number of sampled individuals. Dashes indicate loci that were monomorphic in all sampled populations of a species.

n = 325 n = 110 n = 214

Locus Fis Fit FsT Fis Fit Fst Fis Fit FsT

AAT-1 ——- 1.000 1.000 0.317

AAT-2 0.908 0.961 0.578 1.000 1.000 0.111 0.954 0.984 0.653

ADH-1 1.000 1.000 0.177 1.000 1.000 0.771 1.000 1.000 0.299

GPD 1.000 1.000 0.201 1.000 1.000 0.912

IDH-1 0.863 0.913 0.370 0.760 0.955 0.814 1.000 1.000 0.350

MDH-1 1.000 1.000 0.191 1.000 1.000 0.183 1.000 1.000 0.459

MDH-4 1.000 1.000 0.143 —- """"

6PGD-2 ---- 1.000 1.000 0.089

PGI-1 1.000 1.000 0.144 ----

PGI-2 0.896 0.913 0.166 1.000 1.000 0.777 0.870 0.921 0.393

PGM-1 1.000 1.000 0.193 1.000 1.000 0.368 1.000 1.000 0.870

SOD 0.965 0.991 0.748 ------

TPI-1 ------——

TPI-2 0.929 0.952 0.321 1.000 1.000 0.214 1.000 1.000 0.826

Mean 0.928 0.959 0.430 0.986 0.995 0.640 0.955 0.986 0.694

51 Table 7. Intra- and interspecific seed production in three species of T/wttn/ZantAwf following controlled pollinations.

The mean number of seeds produced per capsule is given, followed by the standard deviation in parentheses; n = number of capsules

examined per treatment. Seed production varied significantly among all treatments within each species except self hand pollination

and intraspecific hand pollination. Individuals ofÆ yZoridhmty produced significantly fewer seeds per capsule for all treatments

coinpared to individuals of 7/. and Æ texanws.

Pollen source for hand pollinations

Outcrossing treatments

Pollen recipient Cleistogamous Chasmogamous Emasculated Self pollen TV. texanus untreated untreated LA K)

148.7 (39.0) 183.1 (36.6) 0.0 130.7 (33.6) 120.9 (34.8) 0.0 0.0

n=100 n- 100 n = 50 n = 100 n = 75 n = 30 n = 35

N. floridanus 107.5 (34.0) 128.3 (28.6) 0.0 96.1 (31.6) 0.0 97.2 (33.2) 0.0

n = 100 n = 100 n = 50 n - 100 n = 50 n = 75 n = 50

157(41.0) 191.1 (46.5) 0.0 138.3 (47.2) 0.0 0.0 125.7 (45.9)

n - 100 n = 100 n = 50 n = 100 n = 45 n = 30 n = 80 Figure 1. Numbered collection localities for 60 populations representing three species of

Nuttallanthus examined for isozyme variation. Collection data are listed in Table 1. Filled symbols (A # 4 ) refer to populations included in population-level isozyme analyses; hollow symbols (AOO) refer to populations sampled but not included in population-level isozyme analyses due to small sample size. Population 58 was divided into two subpopulations (58a and 58b) for analysis.

53 200 km 100 km

+ 41 39

see detau + above

250 km Figure 2. UPGMA phenogram derived from a matrix of Nei's (1972) genetic identity coefficients among 50 populations of three species of Nuttallanthus. Population numbers are listed in Table 1; population localities are depicted in Figure 1. The abbreviations tex, can and flor refer to populations of N. texanus, N. canadensis and N. floridanus, respectively, and are followed by the standard two-letter abbreviation for the state of collection. The cophenetic correlation is 0.920.

55 57 tel OK 58* tel OK d 56 tei OK 59 tex OK 37 tex FL 50 tex FL r £ 47 tex FL 51 tex FL 60 tex SC Group 1 52 tex GA 38 tex AL 62 tex TX 42 tex CA 43 tex CA 40 tex AR 39 tex AR 61 tex TX Group 2 53 tex LA 54 tex LA 58b tex OK 24 can NC 23 can NC 22 can NC 21 can NC 28 can VA 15 can GA 27 can VA 26 can VA - i 19 can MD 18 can MD 05 can DE 01 can FL 06 can FL 08 can FL 12 can FL 11 can FL 17 can GA 25 can SC 09 can FL 20 can MA 14 can GA 16 can GA 32 nor FL 33 nor FL 34 flor FL 35 flor FL 30 flor AL 29 flor FL ■dî31 flor FL 36 flor FL

r i ' ' ' I ' I I I I ' ' ' ' I T— r 1 0.56 0.67 0.78 0.89 1.00 Nei's (1972) Genetic Identity

56 Figure 3. Three-dimensional plot of non-metric multidimensional scaling (MDS) analysis of Nei’s (1972) genetic identity coefficients among 50 populations of three Nuttallanthus species. Population numbers are listed in Table 1; population localities are depicted in

Figure 1. Final stress = 0.122.

57 Æ cfwoikfWK # MDS Axis i n 42/43

oo 29/30 58a

$15

$ 1:^

58b

MDS Ads n Appendix 1. Diagrams of composite banding patterns of all alleles of loci scored for use in an analysis of genetic variation in three Nuttallanthus species. Banding patterns for

MDH-1 and MDH-2 overlapped too extensively for separate loci and alleles to be reliably distinguished; MDH-1 and MDH-2 were treated collectively as a single polymorphic locus with patterns A, B and C corresponding to alleles. was monomorphic in all individuals examined.

59 u 1 1

" 1 1 1 1 1 1 . g ca ^ u T3 ca ^ u

< 1 1 1 1 III^ 1 1 1 1 III& V J \ ^ > V J \ J r-l dy fS m 'f ^ CN

K) ^ O "O ca ^

n i l a 1 1 1

n) u 03 ^ O Xt (D M

1 1 1 § m i l I

(d JD U ca ^ u ca rO u T3

III % III 1 1 1 1 §

r-4 ÇSJ

(Ü ;0 ca fi Ü T) u ^ Q

II 1 1 1 1 1 < III 1

I—1

60 Appendix 2. Genotype data for loci from 60 populations of Nuttallanthus canadensis, N. floridanus and N. texanus. Population numbers refer to Table 1; those followed by an asterisk (*) are included in population-level isozyme analyses.

Locus/ genotype ______1* 2 3 4 5* 6* 7 8* 9* AAT-1 aa 16 5 10 2 30 25 8 19 13

AAT-2 aa -—-— 1 3 ——— 14 ——— ——— -——. ab ——— -——— — 2 ----—™

bb 16 4 7 2 14 24 8 19 13

ADH-1 aa 4 ------3 1 3 2 bb 12 5 7 1 30 25 8 16 11

GPD aa 2 ------2 3 2 2 4 bb 14 5 8 2 30 22 6 17 8

IDH-1

bb 10 5 8 2 12 22 8 19 13

dd 2 ...... -...... MDH-1

bb 13 5 10 2 30 22 8 18 13 CO 3 3 1 ------MDH-4

bb 16 5 10 2 30 25 8 19 13

61 Appendix 2 cont.

Locus/ genotype 1* 2 3 4 5* 6* 7 8* 9* 6PGD-2

CG 16 5 10 2 30 25 8 19 13 PGI-1

bb 16 5 10 2 30 25 8 19 10

PGI-2 uu 4 2 8 2 2 3 ab ...... -...... 1 1 ------bb 12 5 8 2 21 23 8 16 10 be ...... —...... ------

PGM-1

CG 16 5 10 2 30 20 8 19 13

SOD aa 16 1 4 2 29 25 8 18 13 ab —'— ------■—'— 1 ------—— ______bb ------4 6 1 ______TPI-1

dd 16 5 10 2 30 25 8 19 13 TPI-2

bb 9 5 9 1 30 13 5 7 8 be 2 2 ----- cc 7 1 1 10 3 10 5

62 Appendix 2 cont.

Locus/ genotype ______10 11* 12* 13 14* 15* 16* 17* 18* AAT-1 aa 3 10 10 5 20 10 10 10 10

AAT-2

bb 3 10 10 5 20 7 10 10 1

ADH-1

bb 3 10 10 5 20 10 10 10 10

GPD

bb 3 10 10 5 20 10 10 10 10

IDH-1

bb 3 9 10 5 18 10 2 10 6

MDH-1

bb 3 10 10 4 20 10 10 10 10

MDH-4

bb 3 10 10 5 20 10 10 10 10

63 Appendix 2 cont.

Locus/ genotype______10 11* 12* 13 14* 15* 16* 18* 6PGD-2

cc 3 10 10 5 20 10 10 10 10 PGLl

bb 3 10 10 5 19 10 10 10 9

PGL2

ab 3 -...... — bb 3 10 10 5 13 9 10 7 10

PGM-1

cc 3 10 10 5 20 10 10 10 10

SOD aa 3 10 10 5 20 1 10 10 10

TPI-1 aa ......

dd 3 10 10 5 20 10 10 10 10 TTL2

bb 3 6 5 1 9 6 6 9

CC 3 6 4 — ^ % 4 4 1

64 Appendix 2 cont.

Locus/ genotype ______19* 20* 21* 22* 23* 24* 25* 26* 27* AAT-1 aa 17 15 10 10 10 20 10 20 20 yy ______AAT-2 aa 11 33 18 14

bb 5 15 10 10 7 16 10 2 6

ADH-1

bb 17 15 10 10 10 20 10 20 20

GPD

bb 17 15 10 10 10 20 10 20 20

IDH4

14 15 10 10 8 20 10 20 20

MDH-1

17 15 10 10 10 20 7 20 20

MDH-4

bb 17 15 10 10 10 20 10 20 20

65 Appendix 2 cont.

Locus/ ______19* 20* 21* 22* 23* 24* 25* 26* 27* 6PGD-2

cc 17 15 10 10 10 20 10 20 20 PGI-1

bb 15 15 10 10 10 20 10 20 20

PGI-2 aa 1 1 2 2 11 1 1

bb 16 15 7 7 8 9 10 17 17

PGM-1

cc 17 15 10 10 10 20 10 20 20 dd ...... - - ......

SOD aa 11 15 4 8 11 16 ab — ————- —---- ——— ——-——— ------—— 1 bb 6 10 10 6 20 2 9 3 TPI-1

8 ,9. - — — — — ——— — — — — — » — - — — ______— «.M.

dd 17 15 10 10 10 20 10 20 20 TPI-2

88 ——— ______—______bb 17 15 10 10 7 9 6 19 14

66 Appendix 2 cont.

Locus/ genotype ______28* 29* 30* 31* 32* 33* 34* 35* 36* AAT-1 aa 10 11 11 25 10 8 20 10 15

AAT-2

ab ...... -...... ac -...... bb 10 1 ------cc ------11 11 25 10 7 20 10 15

ADH-1

bb 10 9 7 20 4 ------cc ------11 11 25 1 1 6 15 GPD

bb 10 11 11 25 10 8 20 10 15

IDH-1

bb 10 11 11 25 10 8 20 10 2 bd ------......

dd ...... 12 MDH-I aa 3 2 bb 10 11 11 25 7 8 19 10 13

MDH^ aa ...... 4 ...... bb 10 11 11 21 10 8 20 10 15

67 Appendix 2 cont.

Locus/ genotype ______28* 29* 30* 31* 32* 33* 34* 35* 36* 6PGD-2

bb ------11 11 25 9 8 20 10 15 cc 10 — —' -----— ——— ------——— — PGLl

1* 10 11 11 25 10 8 20 10 15

PGL2

bb 10 ————— ———— ——— ————— ———— ————— ———— 13

cc ------13 2 8 20 10 2 dd ------11 11 12 8 — '— ------PGM-1

dd ------11 11 15 10 8 20 10 15 ee ------SOD

bb 10 11 n 25 10 8 20 10 15 TPLl aa ...... -...... 1& — ____

cc ---- — 11 11 25 10 8 20 10 15

TPI-2

bb 10 2 2 ------

cc ------11 11 25 10 5 18 10 15

68 Appendix 2 cont.

Locus/ genotype ______37* 38* 39* jK)* 41 42* 43* 45 46 AAT-1 aa 14 10 10 10 5 5 5 4 1

AAT-2 2 —————— 10 10 3 ————— ——————— ————— ————

cc 12 9 2 5 5 4 1

ADH-1

bb 14 10 10 10 5 5 5 3 1

GPD

as ——— — — 10 10 1 - — bb 14 10 4 5 5 4 1

IDH4

bb 13 10 10 10 5 5 5 4 1

MDH-1 aa 6 10 -...... 2 1 bb 8 10 10 5 5 5 2 ------

MDH^

bb 14 10 10 10 5 5 5 4 1

69 Appendix 2 cont.

Locus/ genotype ______3 ^ 38* 39* 40* 41 42* 46 45 6PGD-2 aa 14 10 10 10 5 5 5 4 1

PGLl

I* 14 m 10 10 5 5 5 4 1

PGL2

bb 11 10 10 9 4 5 5 2 1

PGM-1 aa ------——— 10 10 5 ------

cc 14 10 —- ———------—---- zi I

SOD

bb 14 m 10 10 5 5 5 4 1 IPLl

M 11 m 10 10 5 2 1

TTL2 aa 3 10 10 5 ------bb 3 1 ------—- 5 5 1

cc 8 9 3 1

70 Appendix 2 cont.

Locus/ genotype ______47* /W "W 50* 51* 52* 53* 54* 56* AAT-1 aa 13 5 1 13 7 5 10 12 10

AAT-2 aa ------—-— —------——— —— 2 8 ------

cc 13 5 1 13 7 5 7 4 10

ADH-1

bb 13 5 1 10 7 5 10 12 10

GPD aa ————— — — — ————— ————— ———— ————— 10 12 ———— bb 13 5 1 13 7 5 — ...... — 10

IDH4

bb 13 5 1 13 7 5 10 7 10

MDH-1 3,3. 8 1 .™ .-» —_____ ^ bb 5 4 1 9 6 5 8 11 1

CC ______2 2 __ _ MDH-4

bb 13 5 1 13 7 5 10 12 10

71 Appendix 2 cont.

Locus/ genotype 47* "W; 50* 51* 52* 53* 54* 56* 6PGD-2 M U 5 1 13 7 5 10 12 10

PGI-1

bb 13 5 1 13 7 5 10 12 10

PGI-2

bb 12 4 1 13 2 5 10 9 2

cc 1 ——— ——— —- — 2 ——i-

PGM-1

cc 13 5 1 13 7 5 4 9 10

SOD aa ...... — ......

bb 13 5 1 13 7 5 10 12 10 TPI-1

bb 7 5 1 7 7 5 10 12 10

dd ...... -...... -...... -...... TPI-2 aa 1 10 12 ------bb ———— 3 3 ———— 10

cc 13 1 1 10 7 5 ------

72 Appendix 2 cont.

Locus/ genotype 57* 5&^ 5!W)* 59* 60* 61* 62* AAT-1 aa 12 10 12 10 10 W 15 bb 8 3 ------AAT-2 aa — — 2 ——— 4 — .

cc 20 11 12 6 10 9 15

ADH-1

bb 20 10 12 5 10 10 15

GPD aa ------3 12 4 10 15 bb 20 10 6 10 —

IDH-1

bb 20 13 12 10 10 10 15 bd ...... -......

MDH-1 aa 10 5 d bb 10 8 8 6 10 10 9 cc - ...... 4 4 ...... MDH-4

bb 20 13 12 10 10 10 15

73 Appendix 2 cont.

Locus/ genotype 57* 58a* 58b* 59* 60* 61* 62* 6PGD-2 aa 20 13 12 10 10 10 15 bb cc ------PGI-1 aa bb 20 13 12 10 10 10 15 cc ______PGI-2 aa 4 ab 2 bb 14 12 5 9 10 be ______cc ______dd ______PGM-1 aa 12 4 10 ----- bb cc 20 13 6 10 15 dd ______ee SOD aa ______ab bb 20 13 12 10 10 10 15 TPLl aa bb 20 13 1 6 9 10 15 be 2 ------cc 9 4 1 ...... dd TPI-2 aa bb 19 be cc 1 1 10 9

74 Appendix 3. Summary allele frequencies for 14 putative polymorphic loci within three species of Nuttallanthus. n = number of sampled individuals.

Locus/allele iV. At. feCGMMJ n = 325 n=110 n = 214

AAT-1 a LOOO 1.000 (1949 b 0.051 AAT-2 a (1235 (1180 b (1762 (1009 c &991 0311 d (1009 e (1003 ADH-1 a (1028 0TW7 b (1972 0364 (1949 0 0.636 (1005 GPD a (1034 (1402 b (1963 1.000 (1598 c (1003 IDH-1 a (1082 b (1863 (1886 (1972 c (1019 (1028 d (1037 (1114 MDH-1 a (1045 (1276 b (1969 (1945 (1673 c 0.031 (1009 (1051

75 Appendix 3 cont.

MDH-4 a 0.036 b 1.000 0.964 1.000 6PGD-2 a ---- 0.009 1.000 b ---- 0.991 c IXWO ---- PGI-1 a 0/W6 ---- b 0.978 1.000 LOOO c 0.015 PGI-2 a 0J^2 0385 b 0.835 0318 0799 c 0.012 0300 0.017 d 0382 PGM-1 a ------0257 b 0.047 c 0.985 (1091 0696 d ---- 0.909 e 0.015 ---- SOD a 0J32 b &268 1.000 LOOO TPLd a ---- 0XW7 b 0818 c 1.000 0136

d 1.000 """" -—- TPL2 a 0.009 0332 b OJOO 0.036 0327 c 0300 0.955 0341

76 Appendix 4. Wright's fixation indices (F ) for 42 populations o f three species, indicating deviation 6om Hardy- Weinburg genotypic expectations. SigniScance levels were calculated by pooling genotypes and comparing the frequencies obtained with those expected under random mating. Dashes indicate monomorphic loci. Populations that were monomorphic across aU loci are not represented in the table. Population numbers refer to Table 1. *P<0.05; **P<0.01; ***P<0.001

Populations Locus 1 5 6 8 9 11 12 14 15 16 17

AAT-1 ------— ------:------AAT-2 0.867*** 1.000*** ------1.000** ------ADH-1 1.000*** ------1.000*** 1.000*** ------GPD 1.000*** 1.000*** 1.000*** 1.000*** ------— ------IDH-1 0.877** 1.000*** 1.000*** ------— 1.000** ------1.000*** — 0.341 MDH-1 1.000*** 1.000*** LOOO*** — ------— ------MDH-4 ------— ------— — ------6PGD-2 ------— ------PGI-1 ------— LOOO*** ------1.000*** ------PGI-2 1.000*** 0.918*** 1.000*** 0.770** 1.000*** ------0.624** 1.000** ------1.000** PGM-1 ------1.000*** ------— ------— ------— SOD -0.017 1.000*** ------1.000** ------TPI-1 ------— --- — --- —------TPI-2 1.000*** (1838*** 0.784** LOOO*** 0.780* 1.000** 1.000*** 1.000** 1.000** 1.000** Appendix 4 cont.

Populations Locus 18 19 21 22 23 24 25 26 27 31 32

AAT-1 - — “ “ “ ------— ------—— ------—— :------AAT-2 0.608 0.866*** — ------1.000** 0.827*** ------1.000*** 1.000*** ------ADH-1 ------1.000** LrrjJ IDH-1 1.000** 0.766*** — ------1.000** ------MDH-1 ------1.000** ------1.000** 00 MDH-4 ------— ------1.000*** —

------— — 6PGD-2 “““ 1.000**

------—— — PGI-1 1.000** 1.000*** — PGI-2 1.000*** 1.000** 0.733* 1.000** 1.000*** ------0.779** 0.444* 1.000*** 1.000** PGM-1 ------— ------— ------1.000*** ------SOD 1.000*** — ------1.000** ------1.000** 1.000*** 0.827*** — ------TPI-1 ------— ------— —— TPI-2 1.000** ------0.733* 1.000*** 1.000** 1.000*** 0.875*** ------Appendix 4 cont.

Populations Locus 33 34 35 36 37 38 40 47 50 51 53

AAT-1

AAT-2 1.000** 1.000 * * * 1.000 ** 0.733^

ADH-1 1.000 ** 1.000 ** 1.000*** GPD IDH-1 0.760** 1.000***

MDH-1 1.000*** 1.000 ** * 1.000*** 1.000* * * 1.000* * * 1.000* * 1.000** MDH-4 6PGD-2 PGI-1

PGI-2 1.000 * * * 1.000*** 1.000 * * 1.000 *** 0.689 PGM-1 1.000** SOD

TPI-1 1.000*** 1.000*** 0.675*

TPI-2 1.000* * 1.000*** 1.000* * * 1.000 ** 1.000*** Appendix 4 cont.

Populations Locus 54 56 57 58a 58b 59 60 61 62

AAT-1 ------1.000*** 1.000*** ------AAT-2 1.000** ------1.000*** ------1.000** 1.000** ------

------— ------ADH-1 1.000*** 1.000** —

GPD ------1.000*** ------1.000** ------'------IDH-1 1.000** ------— ------

00 ------o MDH-1 1.000** 1.000** 1.000*** 1.000*** 1.000** 1.000** 1.000*** MDH-4 ------— ------— ------—

6PGD-2 — —— ------— PGI-1 ——— ------— ------PGI-2 0.400 1.000** 0.733** 0.839** ------1.000** 1.000** — 1.000*** PGM-1 1.000** — ------1.000** — ------

o iJ U TPI-1 ------0.400 1.000** 1.000** ------

------TPI-2 1.000*** 1.000*** 1.000** — 1.000*** CHAPTER 2.

MORPHOLOGICAL VARIATION AND SYSTEMATICS OF

NORTH AMERICAN SPECIES OF (LAMIALES)

81 ABSTRACT

The genus in North America consists of three currently recognized species of annual or hiennial herbs with heteromorphic stems and showy personate and bilabiate flowers. High levels of intraspecific morphological variation and intergradation in the morphological characters used to distinguish among these taxa have led to diverse taxonomic treatments of the members of this group. To assesspatterns of morphological variation within and among these taxa and to delineate morphological gaps within the genus, 48 vegetative, floral, fruit and seed characters were measured from field-collected samples acquired from 50 populations of Nuttallanthus. Univariate analyses of morphometric data indicated significant variation among taxa in all of the measured traits. Principal component analysis and cluster analysis using individuals and population means as OTUs consistently separated samples into three primary clusters corresponding to the currently circumscribed species; reproductive characters provided much better discrimination than did vegetative characters. Individuals of the three taxa were correctly classified by discriminant function analysis in 79.8% of cases using only vegetative traits and in 100% of cases using only reproductive characters and all characters. Correlation analysis indicated positive and statistically-significant associations between phenotypic distance and linear distance between populations of each species. These results were consistent with analyses of genetic variation in the North American species of

Nuttallanthus’, strong and significant correlations between phenotypic distance and genetic distance were observed among populations and species. Chromosome counts were obtained from each population of the three species represented in this study; all counts were n - 6. The observed morphological discontinuities separating groups in this

82 genus and the close correspondence of extensive morphological and genetic gaps among these taxa strongly support the recognition of three species in North American

83 INTRODUCTION

Three native North American species of annual or biennial herbaceous plants are ,

currently recognized in the genus Nuttallanthus: N. canadensis (L.) D. A. Sutton, N.

/ZonWaMWj' (Chapman) D. A. Sutton and A. t&ca/nty (Scheele) D. A. Sutton (Sutton 1988,

USDA/NRCS 2002). ). Awtm/ZaMtAwa is native to much of North America

and is naturalized in temperate regions of South America and in Europe, where it has been cultivated as an ornamental for its showy, fragrant flowers (Tutin et al. 1972). N.

is native to the southern United States and Mexico, may be native to South

America, and is naturalized in other temperate areas (Sutton 1988). Nuttallanthus floridanus is more narrowly distributed and occurs in the Atlantic and Gulf coastal plain

of a few states in the southeastern U.S.A. Species ranges overlap in southeastern North

America where all three taxa occur occasionally in dense, mixed populations. These

species are commonly found in sandy soils of open coniferous woodlands, dunes and

crop fields; they may occur in a wide variety of subtrates as adventives in heavily

disturbed sites. All species of Nuttallanthus produce erect fertile stems with simple,

linear, alternately-arranged leaves, and sterile prostrate stems with elliptical, verticillate

leaves. The strongly zygomorphic, bilabiate flowers are borne on terminal racemes and

produce oblong-ovoid bilocular capsules containing numerous radially-symmetrical

seeds. Nuttallanthus canadensis and N. texanus produce flowers with long, slender spurs

at the anterior base of the corolla and have been segregated on the basis of flower size

and seed coat features; the flowers of A. are relatively small (8 to 11 mm in

length, exclusive of the spur) and the surfaces of its seeds possess narrow, longitudinal

ridges separating smooth faces, whereas individuals of A. texanzz.; produce larger flowers

84 (14 to 22 mm) and moderately- to densely-tuberculate seeds lacking longitudinal ridges.

In both species the flowers are generally greater in length than the subtending pedicels.

The inflorescence axes, pedicels and sepals of individuals of N. texanus are glabrous, whereas those of N. canadensis are sparsely to moderately glandular pubescent.

M/ttaZ/antAw.; yZoncZo/m.; has traditionally been distinguished from its congeners by the relative length of the pedicels, which greatly exceed the corolla in length, and the small size (5 to 7 mm in length) of the essentially spur-less flowers. In contrast to the straight inflorescence axes of its congeners, those of N. floridanus are highly flexuous, with the axes zig-zagging at each flower-bearing node. The inflorescence axes, pedicels and sepals of N. floridanus are densely glandular pubescent, and the seeds are characterized by narrow longitudinal ridges separating faces with short, acute ridges and scattered tubercles.

Pennell (1935) proposed that different pollination syndromes explained the variation in floral morphology among species. Our breeding system studies indicate that species of Nuttallanthus are primarily autogamous and that pollinators are rarely successful in transferring pollen among individuals (Crawford 2003). The three North

American species are cross-incompatible, but the nature of the post-pollination reproductive isolating mechanism was not determined.

A chromosome base number of x = 6 has been reported for Linaria and for

Nuttallanthus (Elisens 1986). Previous chromosome counts have been determined for N. caMaùZeAWfx and A. fexonwg. Whereas all counts for A. represent the diploid condition (2n == 12; Kapoor et al. 1987), evidence of polyploidy has been reported in A. texanus, with meiotic chromosome counts of n = 6 (Ward 1983a, Ward 1983b) and n =

85 12 (tetraploid; Raven 1963). No chromosome counts have been reported for AA yZonWaMity.

While has been recognized as a distinct species since its

original description as ZmunuyZonWonu (Chapman 1860), AA and AA.

have been accorded specific status by some authors (Pennell 1935, Sutton 1988) and

relegated to varietal rank by others (Pennell 1920, Munz 1926, Rothmaler 1954,

Cronquist et al. 1984). These different taxonomic treatments reportedly reflect high

levels of variation and intergradation among morphological characters used to distinguish

among these taxa (Pennell 1935, Sutton 1988); the intergradation among morphological

characters has been attributed to ecological factors and hybridization among species.

This study employed univariate and multivariate analyses of vegetative and

reproductive traits measured from field-collected specimens to document the extent and

pattern of morphological variation within and among the North American species of

Nuttallanthus. These analyses were performed subsequent to a study of genetic variation

and mating systems in the genus (Crawford 2003). The principal goals of this

investigation were to determine the extent of morphological differentiation within and

among species in AAwiioZ/oMiAwf, to compare the pattern of morphological variation in the

genus to hypothesized species boundaries, to determine the degree of correspondence

between morphological variation and genetic variation inferred from isozymes, and to

evaluate the utility of the morphological characters historically used to delimit these taxa.

An additional aim was to determine chromosome numbers for populations of these taxa

and to investigate the role of reported differences in chromosome number in the lack of

interfertility among these species.

86 MATERIALS A]\D METHODS

MofpAo/ogy. Individual flowering plants were collected 6om throughout each of the sampled populations and pressed for later examination; where possible, plants possessing both flowers and mature fruits were collected. A total of 242 individuals from

59 populations representing 3 species of Nuttallanthus were selected for examination of morphological variation; 113 individuals from 28 populations of A. canadensis, 37 individuals 6om 8 populations of A. yZonWa/nty, and 92 individuals hrom 23 populations of A (Table 1; Figure 1). All of the specimens examined were deposited at the

Bebb Herbarium of the University of Oklahoma (OKL).

Characters for study were chosen on the basis of prior taxonomic treatments of the Nuttallanthus species complex and examination of herbarium specimens and plants in the field. A total of 48 characters was recorded, including 11 vegetative and 37 reproductive traits; of these, 26 were quantitative traits, 14were ratios of those quantitative traits, two were meristic characters and six were qualitative features (Table

2). Plant height was measured from the top of the root stock to the tip of the tallest fertile stem; stem diameter was measured at the base of the tallest fertile stem. To assess the branching pattern of individual plants, nodal density was defined as the number of leaf- bearing nodes divided by the distance in cm between the base of the longest fertile stem

and the first flower-bearing node on that stem. Leaf characters were taken from the

largest, fully-expanded leaf present on fertile (aerial) and sterile (basal) stems. Floral

characters were obtained from the most recently-opened flower on each individual; bract

characters, flowering pedicel length and vestiture density were taken from the bract and

pedicel subtending that flower. Fruiting pedicel length was taken from the pedicel

87 supporting the fruit used for character analysis. To examine the shape of leaves, bracts, floral structures and fruits, length and width measurements of individual structures were combined as ratios. Seeds were collected from an irregular number of mature capsules on each specimen.

Plant height was measured with a ruler; stem diameter, leaf lengths and leaf widths were measured with a vernier caliper. All other quantitative measurements were

made using a Wild M5 stereo microscope and an ocular micrometer at 6X and 12X.

Seed characters were observed with a Wild M5 stereo microscope at 25X and

5OX. In order to ensure that the seed characters chosen for analysis could be reliably

distinguished at these magnifications, seeds from 10 individuals of each of three species

of Nuttallanthus were examined with the scanning electron microscope. Seeds were

obtained from mature capsules collected from field populations or from herbarium

specimens (Table 1). Individuals were chosen from throughout the North American

range of each species. Representative specimens were affixed to aluminum specimen

stubs with double-sided carbon tape and were sputter-coated for 5 minutes with

gold/palladium in an Anatech LTD Hummer VI plasma sputterer. These samples were

observed with an ETEC Autoscan SEM and photographed with Polaroid 665 film at the

Samuel Roberts Noble Electron Microscopy Laboratory at OU. A total of 150 seeds (5

seeds from each of 10 individuals of three species) was observed with the

SEM.

CAromogowe cowmfs. Meotic tissues obtained &om natural populations and

growth chamber-grown plants (Table 1) were used to determine chromosome numbers.

A chromosome count was obtained from at least one individual for each population of the

88 three species represented in the isozyme study (Crawford 2003). Floral buds were fixed in a 3:1:1 solution of 100% ethanol, chloroform and glacial acetic acid for 24 hours and then washed in three changes of 70% ethanol. The fixed floral tissue was incubated in an alcoholic-carmine stain (Snow 1963) for 48 hours at room temperature, then rinsed and stored in 70% ethanol at 12.5°C. Anther primordia were removed 6om the buds and lightly macerated using needle-point forceps and squashed in 45% acetic acid.

Chromosome counts were made and documented photographically with a Leitz Dialux 20 microscope and Delphi Photoautomat 35 mm camera at lOOOX using Kodak EliteChrome

Tungsten film; all slides are in the possession of the primary author.

AoffffxcaZ Univariate statistical analyses were performed with SPSS for

Windows (ver. 11.5.0; SPSS, Inc. 2002). Descriptive statistics and character correlations were caleulated for each species; normality tests were performed for each character.

Samples departing significantly from normality were log-transformed (continuous variables) or square-root transformed (meristic variables). To compare morphological

characters among populations and species, one-way analysis of variance was performed

on each character that could be normalized and a one-way Kruskal-Wallis test was used

for non-normal samples; post-hoc testing was performed using Fisher’s least significant

difference test.

Multivariate statistical analyses were performed using SPSS for Windows and

NTSYSpc (ver. 2.11c; Rohlf 2002). Ratios were omitted from the multivariate analyses

to avoid inordinate weighting of certain characters.

Principal component analysis (PCA) was performed to examine relationships

among morphological characters and to evaluate the pattern of morphological variation

89 among species. Individual specimens served as operational taxonomic units (OTUs).

Missing data were excluded from the analyses. Separate PC As were performed on three

data sets; 1) reproductive plus vegetative characters, 2) reproductive characters and 3)

vegetative characters. These data sets included both quantitative and meristic characters,

so correlation matrices were used in the PC As. A separate PCA was performed using

population means of reproductive and vegetative characters as OTUs in an attempt to

detect patterns of geographic variation among populations.

Discriminant functions analysis (DFA) was employed to determine whether

species could be reliably distinguished based on the character data. Separate DFAs were

performed on three data sets; 1) reproductive plus vegetative characters, 2) reproductive

characters and 3) vegetative characters. As DFA is not affected by differences in the

scaling of variables (Manly 1994), character data were transformed only if their

distributions were non-normal. Missing data were excluded from the analyses.

Dendrograms were generated with the unweighted pair-group method using

arithmetical averages (UPGMA). In order to reduce potential distortion resulting from

the use of dissimilar scales of measurement in different characters and from differences

in the ranges of character values, all continuous variables were log-transformed; the

meristic variables were square-root transformed. Population means were calculated for

each character data set and matrices of all pair-wise resemblance values among

populations were calculated using average taxonomic distance, Euclidean distance and

product-moment correlation as coefficients of resemblance. Cophenetic correlations

were calculated for each of the dendrograms. The procedures described above were also

used to create dendrograms and summary statistics illustrating species-level relationships.

90 Mantel tests (Mantel 1967) were performed to assess the correlation of average taxonomic distance coefficients calculated 6om the morphological data sets with a genetic distance matrix based on Nei's 1972 genetic distance coefficient calculated 6om isozyme data (Crawford 2003). To assess the degree of association of the phenotypic distance between populations within each species and the geographic location of those populations, Spearman’s rank-order correlation analysis was perfomed with SPSS for

Windows on matrices of average taxonomic distance coefficients and of linear distances

(in kilometers) between populations.

RESULTS

MofyAofogfco/ All of the quantitative and meristic characters assessed in this study varied significantly among the 3 species of Nuttallanthus (Table 3). Plants of AwttuZ/uMtAwj are generally more robust than those of their congeners and the measured vegetativeand reproductive structures arelargest in individuals of this species; the leaf, floral and fruit characters are smallest in N. floridanus, and those of N.

canadensis are intermediate in size. While significant differences in the shapes of these

structures (as assessed by combining length and width measurements of structures as

ratios) were observed among species, the length and width of leaves of both fertile and

sterile stems were correlated with one another. In addition, many of the quantitative

measurements of floral structures were correlated, and their correlation coefficients were

statistically significant. None of the characters were significantly correlated with latitude

or longitude and did not appear to vary clinally within species. Although ranges of some

91 characters overlapped broadly among species, differences existed in many characters that served to demarcate morphologically-distinct taxonomic groups.

Observation of seeds 6om each of three species of with a stereo microscope at 25X and 50X and with the scaiming electron microscope indicated that a number of qualitative differences in seed coat morphology existed between species and that many of those differences could be reliably distinguished at the lower magnifications. Seeds of the three species of Nuttallanthus were similar in size and color, were four- to seven-angled and radially symmetrical. The seeds of Nuttallanthus canadensis were characterized by narrow, entire longitudinal ridges separating interstitial regions with scattered unicellular papillae (Figures 2 through 5). Seeds from individuals of TVwfmZ/antAwj' texanwa' fit into one of two morphological categories: the seeds of individuals from the southeastern United States, California, Oklahoma and central Texas were essentially terete, exhibited little evidence of longitudinal ridges and possessed faces densely covered with acute, multicellulartubercles (Figures 6 and 7); the seeds of individuals from Arkansas, Louisiana, eastern Texas and one population in central

Oklahoma were distinctly angular and possessed faces with low, rounded ridges and scattered multicellular tubercles (Figures 8 and 9). Most of the populations of #.

examined in this study produced only one of these seed types; populations of JVi producing different types of seeds were also distinguishable isozymically (Crawford

2003). One population in central Oklahoma (population 58) was composed of

individuals belonging to both of these groups of M that population was divided

into two sub-populations (58a and 58b) for analysis. No unicellular papillae were

observed on any specimen of N. texanus, but these features were too small to be

92 distinguished with a dissecting microscope. The seeds of yZonWanws' were characterized by narrow, entire longitudinal ridges separating faces with short, acute ridges and scattered multicellular tubercles and unicellular papillae (Figures 10 and 11).

Principal component analysis (PCA) of reproductive and vegetative characters clearly separated individuals of the three species in scatterplots (Figure 12); the first three principal components accounted for 69.1% of the variance in the data set. The first principal component explained 48.0% of that variation, consisting primarily of information from characters related to floral bract, calyx, corolla and fruit size; all of these characters possessed positive eigenvectors, indicating that the first principal component is largely an indicator of the size o f reproductive structures (Appendix 1). A high negative loading was observed for pedicel vestiture density, and the presence or absence of acute longitudinal ridges on seed surfaces also contributed substantially. The second principal component accounted for an additional 12.2% of the variance in the data set and possessed high positive loadings for the pedicel length of flowers and fiuits and for corolla color; the remaining two seed coat characters (presence or absence of obtuse sinuate ridges and multicellular tubercles on seed surfaces) contributed to this component, as did the shape of the inflorescence axis. The third principal component explained 8.89% of the total variance and gave its highest loadings for vegetative characters. The first five principal components possessed eigenvalues greater than 1; considering only those principal components (aAer Legendre et al. 1998, Kachigan 1991)

accounted for only 77.1% of the total variation in the data set. This suggested that some

of the assessed characters varied independently of others and that morphological variation in these species cannot be reduced to a few orthogonal principal component

93 axes. However, PCA of reproductive and vegetative characters clearly distinguished three morphologically distinct groups. Similar results were obtained in PCA of reproductive characters only, with the first three components accounting for 79.8% of the variance in the data set (Figure 13). The analysis of vegetative characters did not clearly separate individuals in a few dimensions; a large degree of overlap was observed among species (Figure 14). The PCA using population means of reproductive and vegetative characters as OTUs clearly separated populations of the three species in scatterplots

(Figure 15); the firstand second principal components accounted for 55.5% and 15.1%, respectively, of the variance in the data set. The third principal component explained only 7.8% of the total variance, but plotting the results in three dimensions illustrated the pronounced phenotypic differentiation among the three species of Nuttallanthus and distinguished two principal groups of N. texanus (Figure 16).

Discriminant function analysis (DFA) of morphological characters indicated that individual plants may be assigned to their respective species with a high degree of accuracy: 100% of individuals were correctly assigned on the basis of all characters and reproductive characters alone (Table 4). DFA of the quantitative and meristic characters in these data sets (omitting qualitative characters) yielded similar results, with a minimum of 98.6% of individuals correctly assigned to species, DFA of vegetative characters alone was less accurate, with predicted group memberships ranging from

71.9% to 85.3% (Table 4).

Dendrograms summarizing the phenotypic similarities among populations and species of AwitaZlaMiAu.; were produced using UPGMA (Sneath et al. 1973) in combination with average taxonomic distance. Euclidean distance and product-moment

94 correlation coefficients based on reproductive and vegetative characters. Regardless of the resemblance coefficient employed, the specimens examined fell into three discrete groups with conspecific populations clustering together. A matrix of average taxonomic distance values resulted in the highest cophenetic correlation among the dendrograms produced (0.925); consequently, that dendrogram is reproduced here (Figure 17). Two discrete clusters of populations of Nuttallanthus texanus are apparent; one cluster is predominately composed of populations from the southeastern United States, California,

Oklahoma and central Texas, whereas populations from Arkansas, Louisiana, eastern

Texas and Oklahoma comprise the second cluster.

The results of the morphological analyses described above were broadly

consistent with those of an analysis of genetic variation in the North American species of

Nuttallanthus (Crawford 2003). Mantel tests revealed a significant correlation between

average taxonomic distance coefficients calculated from the reproductive and vegetative

character data set and Nei's (1972) genetic distance coefficients calculated from isozyme

data (r = 0.749; p = 0.002), as well as between reproductive characters alone and genetic

distance (r - 0.810; p = 0.002) and between genetic distance and vegetative characters

alone (r = 0.164; p = 0.007). Spearman’s rank-order correlation analysis was performed

with SPSS for Windows on matrices of average taxonomic distance coefficients and of

linear distances (in kilometers) between populations. These analyses indicated positive

and statistically-significant associations between phenotypic distance and linear distance

between populations for each species: AwiioZ/oMtAwj' canoden.yij' (rg = 0.372, p < 0.001),

A. (rg = 0.493, p = 0.008) and AC (rg = 0.246, p = 0.001).

95 CAroMOfowe cowwA. Meiotic tissues were used to obtain chromosome counts from at least one individual for each population of the three species represented in the isozyme study (Crawford 2003). All counts were n = 6.

DISCUSSION

Univariate and multivariate analyses of morphometric data measured from field- collected specimens revealed significant morphological gaps between three principal groups of North American AwfrnZZanfAw.;. These results were consistent with the recent treatment of the genus by Sutton (1988) and supported the recognition of three species in

North America: ZYMZ^aZZaMfAzw ca»a(ZeM.yZ.y, A yZorZcZa/m.y and A In addition, the analyses affirmed the utility of the morphological characters that have been used to distinguish among these taxa; although the ranges of some of the measured characters

overlapped among taxa, all of the quantitative and meristic characters assessed in this

study varied significantly among the 3 species of Nuttallanthus (Table2; Table 3). None

of the characters were significantly correlated with latitude or longitude and did not

appear to vary clinally within species but rather demarcated morphologically-distinct

groups of populations. Discriminant function analysis (DFA) employing all measured

characters associated 100% of individual plants with their respective species, and

principal component analysis (PCA) demonstrated that many reproductive and vegetative

traits contributed to the observed morphological differentiation among taxa. While both

vegetative and reproductive characters varied significantly among taxa and provided high

loadings on principal component axes, reproductive characters provided much better

discrimination than did vegetative characters.

96 Species of may be readily identified on the basis of seed coat exomorphology, corolla, spur and pedicel length and inflorescence axis shape, characters traditionally used to discriminate among these taxa and in which little intergradation among species was observed by the primary author. These results were consistent with those of isozyme and breeding system analyses in the North American species of

AwrraZZoMrAwf (Crawford 2003), which demonstrated a high degree of genetic differentiation and reproductive isolation among these species. In addition, a pronounced geographic component in the patterns of genetic and morphological differentiation among populations of these species was evident in the data sets; conspecific populations that were geographicallydistant exhibited greater genetic and morphological divergence relative to geographically-proximate populations, suggestingdecreasing levels of gene

flow with increased physical distance among populations. Genetic and morphological

similarities among conspecific populations were statistically correlated with each other

and with geographic distance across the sampled range of each species of

such congruence in genetic and morphological differentiation generally results from

relatively long periods of divergent evolution among isolated populations (Crawford

1989). Althoughthese species are currently sympatric in a portion of their ranges, the

observed divergence in numerous morphological characters and the accumulation of

unique alleles and substantial allele frequency differences among species of Nuttallanthus

are consistent with a geographic (allopatric) mode of spéciation, in which extrinsic

barriers to gene flow lead to a gradual divergence among isolated sets of populations

(Crawford 1990; Grant 1981). No evidence of interspecific hybridization, polyploidy or

recent divergence among these species was apparent in the data; while phylogenetic

97 relationships among species of Nuttallanthus are unclear (due to similar genetic divergence values among species), no indication of progenitor-derivative relationships among these species was evident (Crawford 2003). The observed intersterility among species of was not due to differences in chromosome number, as all of the populations represented in this study yielded a meiotic count of n = 6. A pattern of morphological, genetic and reproductive differentiation similar to that observed in

Nuttallanthus has been reported in a number of plant genera in which geographical spéciation has been inferred (Elisens et al. 1988; Vanderpool et al. 1991; Warwick et al.

1985). Estimated divergence times among species of based on Nei's

(1987) stepwise mutation rate models (Ie and Iea ) calculated from genetic identities

indicate that these species diverged in the late Miocene and Pliocene (Crawford 2003).

During portions of the Pliocene, the east Gulf Coastal Plain, the current center of

taxonomic diversity of Nuttallanthus, consisted of a number of isolated fragments

(Walker et al. 1987). These fragmented areas of the coastal plain apparently served as

réfugia for a number of plant species (Estill et al. 2001; Sortie et al. 2001) and may have

provided opportunities for spéciation among geographically-isolated progenitor

populations of Nuttallanthus.

Significant levels of character differentiation were also detected among two

groups of populations of A. (Figure 17). These two groups differed significantly

in 17 of the 42 quantitative and meristic characters assessed in this study. Individuals of

group 2 tended to be more robust (differing from group 1 in height, stem diameter and

nodal density) and produced larger flowers, which differed from those of group 1 in

corolla length, tube diameter and spur length. A Mantel test revealed no correlation

98 between average taxonomic distance coefficients calculated from the reproductive character data set and the vegetative character data set (r = 0.056; p == 0.137), suggesting that the observed differences in floral size were not the result of differences in overall plant size. DFA indicated that individual plants of A. rexa/zwj' may be assigned to their

group of origin with a high degree of accuracy: 98.2% of individuals of group 1 and

100% of individuals of group 2 were correctly assigned on the basis of all characters.

DFA of the quantitative and meristic characters in these data sets (omitting qualitative

characters) yielded similar results, with a minimum of 96.4% of individuals correctly

assigned. Whereas most of the assessed differences between groups 1 and 2 in vegetative

and reproductive characters were quantitative in nature, individuals from these groups

differed qualitatively in seed exomorphology (Figures 6 - 9). Some authors have

suggested that relatively robust plants with seeds similar to those of group 2 may be

attributed to hybridization between Nuttallanthus canadensis and N. texanus (Sutton

1988), but the results of our morphological,isozyme and breeding system studies provide

no evidence of a hybrid origin of group 2. All individuals from these two groups shared a

marker allele (6PGD-2a) not found in N. canadensis or N. floridanus, and although

members of both groups exhibited intersterility with individuals of A. canadensis and A

floridanus, no reduction in interfertility was observed between the two groups (Crawford

2003). Considering the minor degree of morphological and genetic differentiation among

these groups (relative to the extent of divergence among species ofNuttallanthus) the

recognition of infraspecific taxa in A. texanus does not seem warranted.

99 LITERATURE CITED

BOWDEN, W. M. 1966. In Love, A. [ed.] lOPB chromosome number reports VIII.

Taxon 15 (7): 279-284.

CRAWFORD, D. J. 1989. Enzyme electrophoresis and plant systematics. In D. SOLTIS

AND P. SOLTIS [eds.], Isozymes in Plant Biology: 146-164. Advances in Plant Sciences

Series, Vol. 4. Dioscorides Press, Portland.

CRAWFORD, D. J. 1990. Plant Molecular Systematics. John Wiley and Sons, New

York.

CRAWFORD, P.T. 2003. Biosystematics of North American species of VwrtaZZuMtAwj

(Lamiales). Ph.D. dissertation, University of Oklahoma: Norman, OK.

CRONQUIST, A., A.H. HOLMGREN, N.H. HOLMGREN, J.L. REVEAL, AND P.K.

HOLMGREN. 1984. Intermountain flora: Vascular plants of the intermountain west,

U.S.A. Vol. 4, The New York Botanical Garden, Bronx, New York.

ELISENS, W. J. 1986. Pollen Morphology and systematic relationships among new

world species in Tribe Antirrhineae (Scrophulariaceae). Amer. J. Bot. 73 (9): 1298-1311.

ELISENS, W. J. AND D. J. CRAWFORD. 1988. Genetic variation and differentiation in

the genus Mabrya (Scrophulariaceae-Antirrhineae): systematic and evolutionary

inferences. American Journal of Botany 75(1); 85-96.

ESTILL, J. C. AND M. B. CRUZ AN. 2001. Phytogeography of rare plant species

endemic to the southeastern United States. Castanea 66 (1-2): 3-23.

GRANT, V. 1981. Plant Spéciation, 2"^ edition. Columbia University Press, New York.

KACHIGAN, S. K. 1991. Multivariate Statistical Analysis, 2°'^ ed. Radius Press, New

York.

100 KAPOOR, B. M., S. RAMCHAEUTAR AND C. GERVAIS. 1987. Liste annotée de nombres chromosomiques de la flore vasculaire dunord-est de l’Amérique. Naturaliste

Canad. 114: 105-116.

KRUSKAL, J. B. 1964. Nonmetric multidimensional scaling: a numerical method.

Psychometrika 29: 28-42.

LEGENDRE, P. AND L. LEGENDRE. 1998. Numerical Ecology, 2"^ English edition.

Elsevier, Amsterdam.

LEWIS, W. H. 1959. Chromosome numbers for Ampélopsis arborea (Vitaceae) and

Linaria texana (Scrophulariaceae). Southwest Nat. 3 (1-4): 214.

MANLY, B. F. 1994. Multivariate Statistical Methods. Chapman and Hall, London.

MANTEL, N. 1967. The detection of disease clustering and a generalized regression

approach. Cancer research 27: 209-220.

MUNZ, P. A. 1926. The Antirrhinoideae-Antirrhineae of the New World. Proc. Calif.

Acad. Sci. 15: 323-397.

NEI, M. 1972. Genetic distance between populations. American Naturalist 106:283-292.

PENNELL, F, W. 1920. Scrophulariaceae of the southeastern United States. Contr. New

York Botanical Garden 221: 224-291, reprinted from Proc. Acad. Nat. Sci. Philadelphia,

1919.

PENNELL, F. W. 1935. The Scrophulariaceae of eastern temperate North America.

Academy of Natural Sciences of Philadelphia, Monograph No. 1: i-xv, 1-650.

RAVEN, P. H. 1963. Amphitropical relationships in the floras of North and South

America. The Quarterly Review of Biology 38 (2): 151-177.

101 ROHLF, F. J. 2002. NTSYSpc: Numerical Taxonomy System, ver. 2.11c. Exeter

Publishing, Ltd.: Setauket, NY.

SNEATH, P. H. AND R. R. SOKAL. 1973. Numerical Taxonomy. WH Freeman, San

Fransisco.

SNOW, R. Alcoholic hydrochloric acid-carmine as a stain for chromosomes in squash preparations. Stain Technology 38:9-13.

SORRIE, B. A. AND A. S. WEAKLEY. 2001. Coastal plain vascular plant endemics: phytogeographic patterns. Castanea 66 (1-2): 50-82.

SPSS FOR WINDOWS ver. 11.5.0. 2002. SPSS, Inc., Chicago, IL.

SUTTON, D A. 1988. A revision of the tribe Anthirrhineae. Oxford Univ. Press, London.

TUTIN, T. G., V. H. HEYWOOD, N. A. BURGES, D. M. MOORE, D. H.

VALENTINE, S. M. WALTERS AND D. A. WEBB [eds.] 1972. Flora Europaea

Volume 3, Diapensiaceae to Myoporaceae. Cambridge University Press, London.

VANDERPOOL, S. S., W. J. ELISENS AND J. R. ESTES. 1991. Pattern, tempo and mode of evolutionary and biogeographicdivergence in Oxystylis and Wislizenia

(Capparaceae). Amer. J. Bot. 78(7): 925-937.

USDA, NRCS. 2002. The PLANTS Database, Version 3.5 (http://plants.usda.gov).

National Plant Data Center, Baton Rouge, LA 70874-4490. U.S.A.

WALKER, H. J. AND J. M. COLEMAN. 1987. Atlantic and Gulf Coastal Province. In

GRAF, W. L. [ed,] Geomorphic Systems of North America. Geological Society of

America, Centennial Special Volume 2: 51-110. Boulder, CO.

WARD, D. E. 1983a. Chromosome counts from New Mexico and southern Colorado.

Phytologia 54: 302-309.

102 WARD, D. E. 1983b. /m LOve, A. [éd.] lOPB chromosome number reports LXXX. Taxon

32 (3): 504-511.

WARWICK, S. I. AND L. D. GOTTLIEB. 1985. Genetic divergence and geographic spéciation in Layia (Compositae). Evolution 39; 1236-1241.

103 Table 1. Numbered collection data for 60 populations representing three species of Nuttallanthus

examined for morphological and chromosome number variation. Population codes are given in

bold numbers; OTUs included in population-level morphological analyses are indicated by an

asterisk (*), Collection numbers for specimens providing seed for environmental chamber-

grown plants used for chromosome number determination are indicated by a plus sign

Voucher collections are deposited at OKL unless otherwise indicated.

Alabama: Baldwin Co.: 1* = 30.414N 87.598W, 0.1 mi. NE of Highway 98 & S Michigan,

Elberta, j jjj, 2 = 30.238N 87.882W, Miller Memorial

Cemetery, Miller Memorial Road and Highway 180 intersection, Crawford 684, 685. Mobile Co.:

3 = 30.433N 88.144W, grounds of Bellingrath Gardens Estate, Crawford 321, 322, 323;4 =

30.243N 88.078, E end of Dauphin island in seaside dunes, 649, 660^.

Delaware: Sussex Co.: 5* = 38.574N 75.056W, 2.0 mi. N of the Highway 1 & Highway 26 intersection, 203, 204^, 206, 206, 207, 20&, 209, 270.

Florida: Bay Co.: 6* = 30.204N 85.847W, 3.0 mi. W of the Highway Alt98 & Highway 392 intersection, Cmn^r<7376, 376, 377, 37g, 379", 770, 777, 772, 773; Calhoun Co.: 7 = 30.464N

85.045W, 1.0 mi. N of Blountstown, <362. Franklin Co.: 8 * = 29.833N 84.876W, 8.6 mi. N of the Highway 65 & Highway 98 intersection, 400, 407\ 402, 403, 404, 406; 9*

= 29.909N 84.3 94W, dunes on Lighthouse Point, W end of peninsula, Crawford 444, 446^, 446,

447, 448; 10 = 29.853N 84.664W, open pine woodland near Carrabelle, N of 3"^ Street, Crawford

307. Lafayette Co.: 11* = 30.139N 83.290W, 0.1 mi. N of Highway 27, 8.8 mi. W of Mayo,

490, 497, 492"^. Wakulla Co.: 12* = 30.136N 84.326W, 0.1 mi. NE of the Highway 98

& Spring Creek Highway intersection, 372, 373, 374, 376, 376^. Walton Co.: 13 =

104 Table 1 cent.

30.309N 86.102W, shore of Eastern Lake, E of Seagrove Beach, Crawford 703, 704, 705.

Georgia: Camden Co.: 14* = 30.759N 81.658W, meadow above N bank of St. Mary's River, 0.1 mi. E of Interstate 95, 503, JO'/, 505, 505, 507, 50g, 517+. Candler Co.: 15* = 32.355N

8L989W, 11.7 mi. SE of the Interstate 16 & Highway 57 intersection, 5P2, 593, 59^/,

595, 60f. Glynn Co.: 16* = 31.020N 81.435W, meadow near St. Andrew picnic area, S

Riverside Drive, Jekyll Island, Crawford 529, 530, 531, 532+. Liberty Co.: 17* = 31.675N

81.414W, edge of Beltowne marsh 3.6 mi. S of Retreat, 5'/3, J'/'/, 5'^5+.

Maryland: Caroline Co.: 18* = 38.817N 75.748W, 1.5 mi. E of the Highway 404 & Highway 16 intersection, Crawford 241, 242, 24f~. 243, Worcester Co.: 19* = 38.096N 75.499W, fallow cornfield E of Highway 113 and 0.3 mi. S of mile 6 marker, Crawford 7&0+, 181, 182, 183, 184,

7&5.

Massachusetts: Middlesex Co.: 20* = 42.504N 71.265W, sandy roadside at edge of pine woodland, Concord Field Station, Bedford, Crawford 898, 899^, 900.

North Carolina: Currituck Co.: 21* = 36.278N 75.915W, 0.5 mi. N of the Highway 158 &

Highway 3E intersection, 775, 775, 777, 779+ . Dare Co.: 22* = 35.261N 75.579W,

Hatteras Island, near intersection of Highway 12 & Paradise Lane, Crawford 101, 102, 103,7 04+.

Duplin Co.: 23* = 34.926N 77.652W, S of Highway 24 and 1 mi. W of Duplin/Onslow county line, Crawford 88, 89, 90+. Hoke Co.: 24* = 35.007N 79.305W, SW of the Highway 211 &

SRI 202 intersection, 52, 53, 54, 55, 55, 57, 53+.

South Carolina: Beaufort Co.: 25* = 32.377N 80.832W, meadow on W bank of Chechessee R,

0.1 mi. SE of Highway 170 bridge, 559, 570, 577, 572, 575+.

105 Table 1 cont.

Virginia: Accomack Co.: 26* = 37.912N 75.356W, Chincoteague Island, S of Highway 175, near causeway to Assateague Island, 75'^, 755, 755, 757, 75^, 759, 770^. Northampton

Co.: 27* = 37.145N 75.967W, 0.4 mi. S of the Highway 13 & Latimer Siding Road intersection,

72g, 729+ 750, 757, 752, 755. Orange Co.: 28* = 38.261N 77.980W, S of Highway 20, near the Highway 20 & Village Road intersection, 25"7, 255, 255, 255+.

Alabama: Baldwin Co.: 29* = 30.238N 87.882W, Miller Memorial Cemetery, Miller Memorial

Road and Highway 180 intersection, 555, 559, 570, 577, 572, 575+. Mobile Co.: 30* =

30.243N 88.078, E end of Dauphin island in seaside dunes, Crawford 552, 555, 654, 555, 555,

557".

Florida: Bay Co.: 31* = 30.204N 85.847W, S ofBeach Front Road, 3.0 mi. W of the Highway

Alt98 & Highway 392 intersection, Cran^r<7 779, 720, 727+ 722. Franklin Co.: 32* = 29.909N

84.394W, dunes on Lighthouse Point, W of Alligator Point village, W end of peninsula,

452+ 455, 454, 455, 455; 33* = 29.723N 84.890W, 0.2 mi. S of the Highway 98 &

Highway 300 intersection, Crmv/ôr(7 745, 749+, 750, 757, 752; 34* = 29.724N 84.899W, sandy meadow SE of E end of Gorrie Bridge, E bank of Apalachicola River, 757, 755, 759,

770, 777"^; 35* = 29.853N 84.664W, N of 3^^ Street in Carrabelle, 792, 795, 794, 795+,

795. Marion County: Rt. 44, 1 mile N of the intersection of Rt. 44 and Rt. 40, W of Astor.

Godfrey 80417 (FSU). Putnam Co.: 36* = 29.623N 81.912W, 0.1 mi. S of Highway 20, W of

Interlachen, FL, 552+ 555, 554, 555, 555.

106 Table 1 cont.

Alabama: Baldwin Co.: 37* = 30.414N 87.598W, 0.1 mi. NE of Highway 98 & S Michigan,

Elberta, 334, 333, 33d, 337, 33g, 342\ Mobile Co.: 38* = 30.243N 88.078, E end of

Dauphin island in seaside dunes, Crawford 636, 637, 638,63f.

Arkansas: Conway Co.: 39* = 35.171N 92.755W, 0.1 mi. S of Interstate 40 mile marker 107,

4P, 30, 37, 32+. Crawford Co.: 40* = 35.528N 94.041W, 0.25 mi. S of Interstate 40, 1.0 mi. W of Mulberry exit, 3d+, 37, 3&; Logan Co.: 41 = 35.300N 93.634W, Subiaco

Academy grounds, Subiaco, 237, 23,3, 23P, 274+.

California: Monterrey Co.: 42* = 36.51 IN 121.942W, grown from seed collected at Point Lobos

State Reserve, P77, P7&, P7P, P07+. Santa Barbara Co.: 43* = 34.044N

119.718W, grown from seed collected at Pelican Bay, Santa Cruz Island, Crawford 926, 927,

P23, lE/jE/üg/w P00+.

Florida: Bay Co.: 44 = 30.204N 85.847W, 3.0 mi. W of the Highway Alt98 & Highway 392 intersection, Crawford 374. Calhoun Co.: 45 = 30.464N 85.045W, 1.0 mi. N of Blountstown,

Crawford 861. Franklin Co.: 47* = 29.909N 84.394W, dunes on Lighthouse Point, W end of peninsula, 423, 427, 423, 42P, 430, 437+; 48 - 29.723N 84.890W, 0.2 mi. S of the

Highway 98 & Highway 300 intersection, Crmt^r(7 737; Wakulla Co.: 50* = 30.136N 84.326W,

0.1 mi. NE of the Highway 98 & Spring Creek Highway intersection, Crawford 477, 478, 479,

Cran^r(7 327, 323, 330+. Walton Co.: 51* = 30.309N 86.102W, shore of Eastern

Lake, E of Seagrove Beach, 3P7, 3P2+, 3P3, 3P4, 3P3.

Georgia: Candler Co.: 52* = 32.355N 81.989W, meadow 11.7 mi. SE of the Interstate 16 &

107 Table 1 cont.

Highway 57 intersection, Crawÿ&fc/ 5^-^, 5^5, 5<96, 5<ÿ7^.

Louisiana: St. Landry Parish: 53* = 30.540N 92.028W, 0.2 mi. E of the Interstate 49 & Highway

190 intersection, Crmtÿôrd 295, 294, 295\ St. Tammany Parish: 54* = 30.298N 89.666W, 0.3

mi. S of Interstate 12, high on W bank of Pearl River, 50(5^, 507, 50&.

North Carolina: Currituck Co.: 55 = 36.278N 75.915W, 0.5 mi. N of the Highway 158 &

Highway 3E intersection, fT 7/4.

Oklahoma: Cleveland Co.: 56* = 35.214N 97.328W, T9N RIW Sec 32 NW 1/4, Crnn^rd 5/,

32, 55, 54, 35, 975, 976, 977, floral buds collected 24 April 2000. Garvin Co.: 57* = 34.708N

97.543W, T3N R3W Sec 30 NE 1/4, 26, 27, 2g, 29, 50, 972, 975, 974, floral buds

collected 23 April 2000; 58* = 34.745N 97.548W, T3N R3W Sec 19 SE 1/4, 954, 955,

936, 957, 938, 959, 940, 941, 942, floral943, buds collected 23 April 2000. Johnston Co.: 59* =

34.327N 96.770W, T3S R5E Sec 3 NE 1/4, "Islands" of shallow soil in Tishomingo granite of

Ten-acre Rock, ^^6, ^<97, 969, 970, 97/, floral buds collected 22 April 2000.

South Carolina: Beaufort Co.: 60* = 32.377N 80.832W, meadow on W bank of Chechessee R,

0.1 mi. SE of Highway 170 bridge, Cmn^rd555, 556, 557, 555, 560^.

Texas: Harrison Co.: 61* = 34.470N 94.595W, 0.1 mi. N of Interstate 20, 0.8 mi. W of exit 604, 250, 25/, 252, 255+. Smith Co.: 62* = 32.469N 95.389W, 0.25 mi. S of

Interstate 20 near mile marker 557, 566, 567, 565+, 569, 570.

108 Table 2. Description of 26 quantitative characters, 14 ratios of quantitative characters, 2 meristic

characters and 6 qualitative characters included in a morphometric analysis of three

Nuttallanthus species. Qualitative character state codings are given in brackets,

Vegetative characters:

1. Plant height (top of the root stock to tallest fertile stem tip)

2. Stem diameter (at base of tallest fertile stem)

3. Number of fertile stems present

4. Number of sterile stems present

5. Nodal density (number of leaf-bearing nodes divided by distance in cm between base of

longest fertile stem and first flower-bearing node on that stem)

6. Sterile stem leaf length (base to tip of blade)

7. Sterile stem leaf width (maximum width)

8. Sterile stem leaf length/width ratio

9. Fertile stem leaf length (base to tip of blade)

10. Fertile stem leaf width (maximum width)

11. Fertile stem leaf length/width ratio

Reproductive characters:

12. Inflorescence axis: straight [1], flexuous [2]

13. Floral bract length (bract subtending the flower used for character analysis)

14. Floral bract width (bract subtending the flower used for character analysis)

15. Floral bract length/width ratio

16. Flowering pedicel length (pedicel supporting the flower used for character analysis)

17. Fruiting pedicel length (pedicel supporting the fruit used for character analysis)

109 Table 2 cont.

18. Pedicel vestiture density: essentially glabrous [1], sparsely pubescent [2], densely

pubescent [3]

19. Adaxial calyx lobe length

20. Adaxial calyx lobe width at base

21. Adaxial calyx lobe length/width ratio

22. Abaxial calyx lobe length

23. Abaxial calyx lobe width at base

24. Abaxial calyx lobe length/width ratio

25. Corolla color (distribution and color of pigmentation): blue with white palate [1], pink with

white palate [2], lavender throughout [3], dark purple throughout [4]

26. Corolla length (tip of abaxial lobe to base of ovary)

27. Adaxial corolla lobe length

28. Adaxial corolla lobe width

29. Adaxial corolla lobe length/width ratio

30. Adaxial corolla sinus depth

31. Adaxial corolla lobe length/sinus depth ratio

32. Abaxial corolla lobe sinus depth

33. Abaxial central corolla lobe width

34. Abaxial corolla sinus depth/lobe width ratio

35. Corolla tube length (base to mouth of tube)

36. Corolla tube diameter (mouth of tube)

37. Corolla tube length/diameter ratio

38. Corolla length/corolla tube length ratio

110 Table 2 cont,

39. Corolla spur length

40. Corolla spur diameter at base

41. Corolla spur length/diameter ratio

42. Corolla spur length/corolla length ratio

43. Capsule length (largest capsule)

44. Capsule diameter (at maximum)

45. Capsule length/diameter ratio

46. Acute, entire longitudinal ridges on seed surface; present [1], absent [2]

47. Low, obtuse, sinuate ridges on seed surface: present [1], absent [2]

48. Acute, discrete multicellular tubercles on seed surface; present [1], absent [2]

111 Table 3. Descriptive statistics for 42 quantitative and meristic morphological characters utilized in morphometric analyses of three species of

AWiaZ/oniAwf. Parentheses denote the number of populations (n) and the number of specimens (N) examined. Means and standard

deviations are provided, as are the results of ANOVA, Kruskal-Wallis and a posteriori tests, with species ordered (left to right) horn low

to high values. Lines above and below species abbreviations connect species that do not differ significantly for a given variable.

Dimensions are in mm unless otherwise noted.

N. canadensis N. floridanus fn = 28.N=1131 rn = 8.N = 3Tl fn = 23.N = 921 ANOVA a posteriori test Character Mean SD Mean SD Mean SD

to 1. Height (cm) 38.9 7.37 29.4 8.04 40.7 8.67 F = 27.7; p < 0.001 yZor can tex

2. Stem width 1.48 0.35 1.17 0.41 1.64 0.40 F = 20.8; p < 0.001 yZor can tcx

3. Fertile stem number 1.88 1.33 2.57 2.17 1.53 0.94 = 9.05; p = 0.001 can yZor

4. Sterile stem number 9.25 5.90 13.0 12.0 8.91 6.27 F = 4.64; p = 0.011 tax can yZnr

5. Nodal density 0.61 0.16 0.76 0.19 0.75 0.18 F = 20.5; p < 0.001 tex /Zor can

6. Sterile leaf length 4.96 1.65 3.83 0.94 6.15 2.07 F = 23.7; p< 0.001 yZor can tax

7. Sterile leaf width 1.79 0.48 1.68 0.55 2.23 0.63 F = 19.5; p< 0.001 yZor can tax

8. Sterile leaf length/width 2.96 1.47 2.39 0.59 2.79 0.70 F = 3.83; p = 0.023 yZor tcx can Table 3 cont.

Æ yZonjoMüÿ fexoMiw fn = 28.N-113'l fn = 8 .N = 37'l fn = 23.N = 921 ANOVA a posteriori test Character Mean SD Mean SD M ean SD

9. Fertile leaf length 20.8 6.19 18.7 6.25 23.0 4.93 F = 8.10; p < 0.001 _/Z07" can tex

10. Fertile leaf width 1.25 0.29 0.81 0.20 1.86 0.57 F = 104; p < 0.001 yZor can tex

11. Fertile leaf length/width 16.9 4.60 23.1 7.72 12.9 2.73 F = 64.9;p < 0.001 tax can yZor

Reproductive:

13. Floral bract length 2.04 0.34 1.59 0.43 2.65 0.49 F = 100; p < 0.001 yZor can tex

14. Floral bract width 0.62 0.10 0.48 0.09 0.96 0.19 = 165; p < 0.001 can tex

15. Floral bract length/width 3.36 0.61 3.32 0.70 2.81 0.48 F = 24.5; p < 0.001 tex yZor can

16. Pedicel length (flower) 2.40 0.68 5.46 0.97 3.60 1.07 F = 154; p < 0.001 can tex yZor

17. Pedicel length (fruit) 3.64 0.92 10.2 1.64 5.25 1.23 F = 414; p < 0.001 can tex yZor

19. Adaxial calyx lobe length 2.25 0.25 1.97 0.25 2.78 0.42 F = 107; p < 0.001 yZar can tex

20. Adaxial calyx lobe width 0.57 0.10 0.52 0.10 0.79 0.16 F = 99.7; p < 0.001 yZor can tex

21. Adaxial lobe length/width 4.01 0.65 3.82 0.56 3.60 0.59 F = 11.4; p < 0.001 tex /Zor can

22. Abaxial calyx lobe length 2.63 0.30 2.17 0.29 3.30 0.42 F = 172; p < 0.001 yZor can tex Table 3 cont.

A'! ^ondanwf Æ texanwf fn = 28.N=1131 rn = 8.N=371 fn = 23.N = 92') ANOVA a posteriori test Character Mean SD M ean SD Mean SD

23. Abaxial calyx lobe width 0.83 0.10 0.69 0.10 1.22 0.15 F = 359; p < 0.001 yZor COM tex

24. Abaxial lobe length/width 3.19 0.39 3.18 0.44 2.74 0.39 F = 36.2; p < 0.001 tex COM

26. Corolla length 8.75 1.31 6.83 1.08 12.3 1.82 F = 227; p < 0.001 yZor can tex

27. Adaxial corolla lobe length 3.20 0.56 2.37 0.51 4.67 0.74 F = 229; p < 0.001 yZor COM tex

28. Adaxial corolla lobe width 1.77 0.41 1.22 0.35 2.07 0.51 F = 49.2; p < 0.001 COM tex

29. Adaxial lobe length/width 1.87 0.35 1.99 0.30 2.31 0.39 F = 38.3; p < 0.001 COM flor tex

30. Adaxial corolla lobe sinus depth 1.93 0.53 0.95 0.40 2.34 0.68 F = 78.2; p < 0.001 yZor COM tex

31. Adaxial lobe length/sinus depth 1.74 0.36 2.81 1.03 2.12 0.57 F = 55.4; p < 0.001 COM tex

32. Abaxial corolla lobe sinus depth 2.62 0.64 1.93 0.39 3.81 0.78 F = 132; p < 0.001 yZor COM tex

3 3 . Abaxial corolla lobe width 2.81 0.74 1.70 0.54 4.01 0.93 F = 126; p < 0.001 _/Zor COM tex

3 4 . Abaxial lobe sinus depth/width 0.96 0.19 1.18 0.22 0.96 0.17 F = 22.3; p < 0.001 can tex yZor

35. Corolla tube length 3.17 0.41 2.49 0.39 3.71 0.40 F = 125; p < 0.001 yZor COM tex

3 6 . Corolla tube diameter at mouth 1.16 0.16 0.96 0.17 1.51 0.23 F = 138; p < 0.001 yZor COM tex Table 3 cont.

A'i yZorw&ZMKy fn = 28.N=1131 (n = 8. N = 37) fn = 23.N = 921 ANOVA a posteriori test Character Mean SD Mean SD Mean SD

37. Corolla tube length/diameter 2.78 0.42 2.64 0.41 2.50 0.40 F = 11.5; p < 0.001 /gr /7or COM

38. Corolla length/tube length 2.77 0.30 2.77 0.36 3.32 0.44 F = 62.2; p < 0.001 COM /Zor tex

39. Corolla spur length 4.38 1.28 0.26 0.18 7.40 1.55 F = 417; p < 0.001 yZor COM tex

40. Corolla spur diameter at base 0.53 0.09 0.30 0.14 0.71 0.14 F = 162; p < 0.001 yZor COM tex

41. Corolla spur lengtb/diameter 8.37 2.27 0.83 0.32 10.5 2.44 F = 241; p < 0.001 yZor COM tex

42. Corolla length/spur length 2.18 0.85 32.4 19.3 1.77 0.78 F = 968; p < 0.001 tax COM yZor

43. Capsule length 3.26 0.32 2.66 0.32 3.88 0.45 F = 147; p < 0.001 yZor COM tex

44. Capsule width 2.96 0.31 2.36 0.28 3.38 0.30 F = 146; p < 0.001 yZor can tex

45. Capsule length/width 1.10 0.06 1.13 0.09 1.15 0.11 F = 6.98; p = 0.001 COM _/7or tex Table 4. Predicted group memberships (as percentages) of individuals of three species of Nuttallanthus based on Discriminant Function Analysis. N = Number of sampled individuals / species.

A. Vegetative and reproductive characters

Species N A. A. yZonüfanwA' A. t&ranu.;

A. canoùfen.szj' 76 100.0 0 0

A. yZonWnnwj' 32 0 100.0 0

A. 71 0 0 100.0

B. Reproductive characters

Species N A canacfenf A. floridanus A rexaMiw

A. j' 87 100.0 0 0

A. yZonWoMWf 37 0 100.0 0

A. texanity 73 0 0 100.0

C. Vegetative characters

Species N A A yZonWanwf A. rexaMWf

A. 102 85J 4.9 9.8

A 32 28.1 71.9 0

A 89 23.6 0 76.4

116 Figure 1. Numbered collection localities for 59 populations representing three species of

Nuttallanthus examined for morphological and chromosome number variation. Collection data are listed in Table 1. Filled symbols ( A # ^ ) refer to populations included in population-level morphological analyses; hollow symbols ( AOO) refer to populations only included in species-level morphological analyses. Population 58 was divided into two subpopulations (58a and 58b) for analysis. Chromosome counts were obtained for all populations represented by filled symbols and for the two populations represented by hollow symbols and indicated by an " (Table 1).

117 I 1 200 km 100 km

oo

A A *41 39

see detail + above

250 km Figures 2-5: Scanning electron micrographs of seeds of caMCKfena'fj'.

Figure 2. Oblique lateral view of an angular seed, hilum to top (Crawford 17). Scale =

100 ^ m.

Figure 3: Expanded view of longitudinal ridge and face (Crawford 17). Scale = 10 ju m.

Figure 4; Oblique view of distal end of seed (Crawford 19). Scale = 10 jU m.

Figure 5: Oblique lateral view of a unicellular papilla (Crawford 19). Scale = 1 /x m.

119 120 Figures 6 - 9: Scanning electron micrographs of seeds o f JVwtro/ZaMtW

Figure 6; Oblique lateral view of seed, hilum to top (Crawford 27). Scale = 100 jtt m.

Figure 7: Expanded view of seed (Crawford 27). Scale - 10 m.

Figure 8: Oblique lateral view of seed, hilum totop (Crawford 23). Scale = 100 [i m.

Figure 9: Expanded view of seed surface (Crawford 23). Scale = 10 ju m.

121 122 Figures 10 and 11: Scanning electron micrographs of seeds oî Nuttallanthus floridanus.

Figure 10: Lateral view of seed, hilum to top (Godfrey 80417). Scale = 100 [x m.

Figure 11 : Oblique view of distal end of seed (Godfrey 80417). Scale = 10 /t m.

123 a

124 Figure 12 and 13. Two-dimensional plots of principal component analysis (PCA) of characters taken from individuals of three Nuttallanthus species.

Figure 12. PCA of 35 reproductive and vegetative characters taken from 179 individuals of three species. The first and second principal components account for

60.2% of the total variation in the data set. See Table 4 for loadings and eigenvalues.

Figure 13. PCA of 26 reproductive characters taken 6om 179 individuals of three

Nuttallanthus species. The first and second principal components account for 73.3% of the total variation in the data set.

125 12 0.58 AC fowiiM A AC canadensis # AC ykrWwwM * 0.29

a2

oI- 0.01 U A A A. a \ "C PLI -OJl A A A / »

" * * * ^ -0.61 -I ------— fA — 1 ------1------r - -1.53 -0.80 -0.07 0.66 1.39 Principal Component 1

0.66 • - r 13 * I*

A ** A*

0J4

* A # ^ A A A © AA- A *A I 0.02 0» & '« g "C

-OJO- AC texanus ▲ AC canadensis • AC floridanus ♦

-0.62 I I @ I -1.44 -0.65 0.14 0.93 1.72 Principal Component 1

126 Figure 14 and 15. Two-dimensional plots of principal component analysis (PCA) of characters taken from individuals of three Nuttallanthus species.

Figure 14. PCA of 9 vegetative characters taken from 179 individuals of three

Nuttallanthus species. The first and second principal components account for 56.1% of the total variation in the data set.

Figure 15. PCA of population means of 35 reproductive and vegetative characters taken from 50 populations of three Nuttallanthus species. The first and second principal components account for 70.6% of the total variation in the data set. Population numbers are listed in Table 1; population localities are depicted in Figure 1.

127 14 1J7T N. texanus A N. canadensis • N. floridanus ♦ 0.80

eS

0.2 2 - UI

-g" . y * * : •* ; Â - A A A -0.36

- 0 .9 4 -1.48 -0.60 0.28 1.17 2.05 Principal Component 1

0.58 20 23# 15 8# .12 ^#19 N. texanus A 25 * 24 %, N. canadensis • N. floridanus ♦ 0.28 N S eB g" o -0.03 U "3 ÿ W M 50 A 3 7 ^ i P 60 ^ B "C 47^ M ^59 ^ B. 57" i"" 58 » -0.33 54 42 36 A A 39 A ^ 58 b 33A 32 34 A A A 29

y A 3 t 35 -0.63 -1.71 -1.03 -0J4 034 1.02 Principal Component 1

128 Figure 16. Three-dimensional plot of principal component analysis (PCA) using population means of 35 reproductive and vegetative characters taken from 50 populations of three Nuttallanthus species. The first three principal components account for 78.3% of the total variation in the data set. Population numbers are listed in Table 1; population localities are depicted in Figure 1.

129 35 JV. A

Principal Component 3 604 52 34 4

624 OJ o

Principal Component 2 Principal Component 1 Figure 17. UPGMA phenogram derived &om a matrix of average taxonomie distance coefficients among 50 populations of three Nuttallanthus species. Population numbers are listed in Table 1; population localities are depicted in Figure 1. The abbreviations tex, can

yZor refer to populations of A. iexoMwa', A. and A respectively, and are followed by the standard abbreviation for the state of collection. The cophenetic correlation is 0.925.

131 01 cam FL 06 can FL rCE 14 cam GA 15 cam GA 27 cam VA 26 cam VA 28 cam VA 16 cam GA 17 cam GA 22 cam NC 05 cam DE 18 cam MD 19 cam MD 21 cam NC 08 cam FL 25 cam SC - d 09 can FL 11 cam FL 12 cam FL 24 cam NC 20 cam MA 23 cam NC 38 tel AL 52 tei GA 60 tex SC 37 tex FL 58a tex OK 56 tex OK Group 1 57 tex OK 42 tex CA 43 tex CA 47 tex FL 50 tex FL 51 tex FL 59 tex OK 62 tex TX 39 tex AR Group 2 40 tex AR 54 tex LA 53 tex LA LÆ61 tex TX 58b tex OK 31 flor FL 34 Oor FL 35 flor FL 29 flor FL 30 flor AL g 32 flor FL -33 flor FL 36 flor FL I I I I —T " 0.56 0.43 0.30 0.17 0.03 Average Taxonomie Distance

132 Appendix 1. Eigenvalues and loadings on the first three component axes from principal component analysis of morphological variation in three species of Nuttallanthus. Descriptions of characters are provided in Table 2.

Principal Component Axis 1 2 3 Eigenvalues 16.8 4.26 3.11 Variance (as percent of total) 48.0 60.1 69.1 Loadings 1. Plant height 0.563 -0.289 0.486 2. Stem diameter 0.531 -0.199 0.618 3. Number of fertile stems -0.268 -0.113 0.605 4. Number of sterile stems -0.152 0.158 0.329 5. Nodal density 0.033 0.484 -0.355 6. Sterile stem leaf length 0.454 -0.045 0.449 7. Sterile stem leaf width 0.445 0.126 0.484 9. Fertile stem leaf length 0.465 -0.090 0.621 10. Fertile stem leaf width 0.750 0.054 0.372 12. Inflorescence axis -0.730 0.627 0.115 13. Floral bract length 0.750 0.067 0.208 14. Floral bract width 0.842 0.114 0.134 16. Flowering pedicel length -0.060 0.837 0.074 17. Fruiting pedicel length -0.380 0.781 0.236 18. Pedicel vestiture density -0.886 -0.069 0.225 19. Adaxial calyx lobe length 0.770 0.133 0.304 20. Adaxial calyx lobe width at base 0.750 0.218 0.108 22. Abaxial calyx lobe length 0.859 0.084 0.226 23. Abaxial calyx lobe width 0.896 0.156 0.004 25. Corolla color 0.578 0.721 -0.033 26. Corolla length 0.898 0.141 -0.231 27. Adaxial corolla lobe length 0.904 0.122 -0.208 28. Adaxial corolla lobe width 0.738 -0.154 -0.162 30. Adaxial corolla sinus depth 0.767 -0.197 -0.208 32. Abaxial corolla lobe sinus depth 0.803 0.126 -0.300 33. Abaxial central corolla lobe width 0.844 -0.002 -0.263 35. Corolla tube length 0.841 -0.062 -0.125 36. Corolla tube diameter 0.849 0.130 0.002 39. Corolla spur length 0.881 -0.152 -0.260 40. Corolla spur diameter 0.805 -0.125 -0.169 43. Capsule length 0.803 -0.022 0.116 44. Capsule diameter 0.826 -0.132 -0.024 46. Acute, entire longitudinal ridges on seed 0.821 0.443 -0.079 47. Low, obtuse, sinuate ridges on seed 0.439 -0.632 -0.317 48. Acute, discrete multicellular tubercles on seed -0.071 -0.840 0.179

133